Facing the truth: Audiences’ facial expressions and emotional responses to real or fake photos in the news

Abstract

Similar to the now-ubiquitous AI-generated photos, multi-purpose stock photos are directly at odds with the values of photojournalism, which seek to visually speak the truth and depict genuine human emotions. Using stock photos and staffed photojournalist photos as stimuli, this study examines effects of truth and emotion in photos in a news context. Facial expression biometric measures, valence perceptions, and arousal levels are the dependent variables. The findings show a number of nuanced interaction effects that suggest overall stronger effects of staff-taken photos. In the contemporary media era of deepfakes and synthetic media, institutions prioritizing human-captured visuals may better safeguard audiences against disinformation, which can weaponize artificial emotional appeals and hallucinate even in innocuous contexts. The present research shows that authenticity is a currency of trust in visual communication, and there is a continued merit to truthful news photographs taken by on-staff photojournalists.

Keywords

Photojournalism ethics truth emotion biometrics stock photos journalism

Introduction

Julianne Newton wrote in 1998 that photojournalism might be “in big trouble, due in part to changing technologies and shifting media economics, but also due to a decline of public belief in visual truth” (pg. 4). A decade later, the question was posed: “In a time when anyone can make anyone appear to do anything in a realistic-looking image, will a profession whose mission is to show the world to itself continue?” (Williams and Newton, 2009: p. 233). At that point, mass photojournalist layoffs had not quite begun, cell-phone photography was just gaining traction, and the contemporary generative technology of visual artificial intelligence was science fiction.

Some of the longstanding fears about the health of the profession have come to fruition. A major decline in the number of professional, on-staff visual journalists within news organizations is very much a reality, influenced by rapidly evolving technology, visual abundance, and the perception of photojournalists as secondary figures (Brennen, 1998; Gade, 2011; Klein-Avraham and Reich, 2016; Nilsson, 2021; Zelizer, 1995). In the fallout, there is sometimes a notion that nearly “any” visual content will suffice (Bock, 2023; Klein-Avraham and Reich, 2016; Sonderman, 2013). Audiences today are faced with these sources as well as photo-realistic AI generated images, usually having a hard time determining fake from real (Wojdynski and Shivers, 2025).

Despite the visual congestion, photojournalists still risk their lives to show the world the images they believe need to be seen. They make a strong argument for the importance of on-staff, professionally-trained photojournalists who follow traditional norms, hold shared values, and profess to maintain ethical standards (e.g. Barnham, 2023). Photojournalists distinguish their images from others’ by purpose: “to show the truth, to the extent any human can discern and communicate, of stories significant to people’s lives” (Williams and Newton, 2009: p. 234). Apturing the spectrum of human emotions is an additional fundamental principle (Lester et al., 2022; Newton, 1998; Rössler et al., 2011). Truth and emotion are at the heart of the present study (Table 1).

Table 1.

Effects of photo emotion and source on self-reported arousal scores.

Factor	Condition	Mean arousal score(M)	Standard deviation (SD)
Emotion	Positive	5.15**	2.10
Emotion	Negative	3.62**	1.94
Source	Staff	4.46	1.90
Source	Stock	4.30	2.39
Interaction	Negative staff	4.07*	1.84
	Negative stock	3.17*	1.95
	Positive staff	4.85	1.91
	Positive stock	5.44**	2.26

*p < 0.05; **p < 0.01.

We examine audience perceptions toward photos taken by professionally-trained, on-staff photojournalists of real newspapers and similar photos from stock image banks. The study uncovers some of the ways in which audiences perceive and place importance upon truth and genuine emotion in photos. As stimuli, we use formulaic and generic stock photographs designed with multiple-purposes and multiple usages in mind (Machin, 2004). We measure if and how viewers respond differently to untruthful images with contrived emotion versus truthful images with candid-captured emotion. The other set of stimuli were award-winning and published photojournalist-shot photographs. Two axioms of photojournalism are the independent variables: truth (as measured by stock photos [untruthful] and staff photojournalist photos [truthful]) and emotion (positive or negative valence) as we are interested in whether views can perceive the absence of these photojournalism principles. The dependent variables include viewers’ facial expressions, self-report valence perceptions, and arousal (Table 2).

Table 2.

Descriptive statistics for valence and arousal scores by emotion and photo source.

Emotion	Photo source	N	Valence					Arousal
Emotion	Photo source	N	M	SD	SE	95%CI (LL)	95% CI (UL)	M	SD	SE	95%CI (LL)	95% CI (UL)
Negative	Stock	41	7.95*	1.22				3.17*	1.95	0.31	2.56	3.79
Negative	Staff	41	7.68*	1.46				4.07*	1.84	0.31	3.46	4.69
Positive	Stock	41	2.49	1.47				5.44	2.26	0.31	4.82	6.05
Positive	Staff	41	2.27	1.16				4.85	1.91	0.31	4.24	5.47

M = Mean; SD = Standard Deviation; SE = Standard Error.

*p < 0.05.

Review of the literature

Photojournalism and truth

Practitioners distinguish photography taken by professional, staffed photojournalists from other information-sharing activities in part by photojournalists’ ethical foundations, values, and norms (Fahmy et al., 2014; Santos Silva & Eldridge II, 2020; Singletary and Lamb, 1984). Photojournalism derives from Western journalism ideals of upholding democracy through an unrestrained flow of information, allowing citizens to govern themselves (Merrill, 1974; Santos Silva & Eldridge II, 2020). Given this mission, truth – “reporting the human experience accurately, honestly, and with an overriding sense of social responsibility” (Newton, 1998, p. x) is at the heart of photojournalism (Elliot, 2011; Kovach and Rosenstiel, 2021; Lester, 2018; Santos Silva & Eldridge II, 2020, p. 31). The National Press Photographer’s Association (NPPA) 2024 Code of Ethics contain emphasizes accuracy, avoiding staging, completeness, and integrity. The first guideline highlights the imperative of accuracy and thoroughness when portraying subjects, while the second emphasizes avoiding contrived or staged photo opportunities. Rule three states the significance of presenting photojournalistic work as completely as possible and with proper contextualization. The fourth rule urges practitioners to steer clear of personal biases, and the fifth cautions against any attempts to manipulate or alter events being documented. Finally, rule six underscores the importance of minimizing photo editing to maintain the integrity of the photographic content and its context (para 11). In the present study, truthful images are defined by candid photographs taken by professional photojournalists.

Photojournalism and emotion

Photojournalists are also trained to capture the depth of human experiences and the fullness of their emotions (Costa, 1950). Photojournalism textbooks stress the importance of conveying high-arousal emotion to engage viewers (e.g., Arthur, 2021; Kobré, 2011; Skinner, 2010; Horton, 1989). Emotional and intimate news photos have long been acknowledged for their ability to evoke emotional responses and prompt action (Silva Santos & Eldridge II, 2020). The portrayal of emotion is a critical criterion in judging photo contests, with emotionally resonant photos nearly always securing more awards (Lough, 2021; Singletary and Lamb, 1984). The preference is echoed by editorial departments, who choose emotional photos for publication more frequently than others (Rössler et al., 2011). Professional photography by on-staff photojournalists generally exhibits more emotion compared to non-professional work (Greenwood and Thomas, 2017; Mortensen and Gade, 2018). Negative emotions, specifically, are most often published and rewarded. Singletary and Lamb’s (1984) seminal examination of award-winning photos from 1978 to 1981 revealed that 81% of the winning entries depicted accidents, disasters, crime, or violence, with only three out of 222 photos being clearly positive. A similar trend was observed in an analysis of photos from 1960 to 2020 (Godulla et al., 2021; Lough, 2021). These findings resonate with foundational work on nonverbal communication, where Friedman (1979) posited that authenticity in emotional expression hinges on the congruence between internal states and outward displays—a principle central to photojournalistic training. Similarly, Tucker and Riggio (1988) demonstrated that spontaneous facial expressions are perceived as more genuine than posed or rehearsed ones, which aligns with photojournalism’s emphasis on candid emotional portrayal.

Research has explored the positive effect of emotional photos upon attention, memory, engagement, beliefs, and behavioral intentions. Emotional photos, particularly those depicting negative events, attract more attention and elicit stronger emotional responses from viewers (Calvo and Lang, 2004; Fenske et al., 2004; Knobloch et al., 2003; Rössler et al., 2011). Previous research indicates that emotional images attract more prolonged attention compared to neutral ones (Calvo and Lang, 2004), demonstrating the influence of emotion on attention (e.g., Eastwood et al., 2001; Fenske and Eastwood, 2003; Fox, Russo, Bowles and Dutton, 2001). Negative images tend to evoke negative responses, whereas positive images elicit positive responses. The intensity of visual content correlates with its impact on emotional responses and news judgment, particularly when eliciting negative emotions (Martin-Kratzer, 2005). Publications featuring negative and alarming images have been noted to capture heightened attention (Knobloch, Hastall, Zillman and Calliston, 2003). Reports emphasizing destruction and violence elicit increased attention, better retention of visual content, favorable evaluations, and heightened interest (Emmett, 2011; Rössler et al., 2011). In research, emotion is defined and measured through two dimensions: valence and arousal (Barrett et al., 2007; Bradley and Lang, 1994). Valence refers to the positivity or negativity of a reaction (Lang et al., 1993), while arousal signifies the level of stimulation (Berger and Milkman, 2012). We adopt this operationalization of emotion.

Photography in the news that is neither truthful nor genuinely emotional

Multi-purpose, orchestrated stock photos directly oppose the photojournalistic values of truth and do not portray real emotion (Aiello, 2016; Thurlow et al., 2020). Companies like Getty Images offer both stock and journalistic imagery, categorized as “creative” or “editorial”. This research focuses on the creative, adaptable style of stock photos. The creative stock photography sector aims to anticipate future needs of various media outlets and create generic images that are easy to use in different contexts so that they can be purchased and used repeatedly (Frosh, 2001). The industry’s nature has given rise to the stereotypical generic stock photo, which has been described as a polished, formulaic, multi-purpose representation (Frosh, 2001: p. 30). Characteristics of “good” stock photos include vibrant colors, precise white balance, sharp focus on the subject against a blurred background, low ISO for clarity, and a horizontal orientation (Wallon-Hárs, 2023). They are considered essential not only for advertising, branding, and publishing but also for journalism (Aiello, 2016; Frosh, 2001). A report on the Global Stock Photos Market noted a revenue of $2 billion in 2020, with an estimated annual growth rate of 7% through 2026 (Expert Market Research, 2021). These kinds of stock photos show neither candid truthful occurrences nor genuine emotion and thus directly defy these two important photojournalistic precepts at the heart of the present study. For the purposes of this research, a stock photo is operationalized as a photo that is created but appears to be a candid moment that might be used in multiple contexts.

While not the stimulus image of the present research, visual generative AI is another relevant type of fake visual that journalism organizations and organizations are having to contend with. Regarding AI images specifically, journalists thus far advocate for a pragmatic and iterative implementation approach (Mothes-et-al, 2024) but express a concern for maintaining trust with the public as awareness of fake imagery may increase audience skepticism (Moran and Shaikh, 2022; Porlezza and Schapals, 2024) and ultimately degrade the credibility of the news industry. All fake images have the potential to misinform audiences, change their memory of the events (Murphy and Flynn, 2022; Nash, 2018) and alter perception of the topic (Bransford and Johnson, 1972; Newman et al., 2015).

Previous research of non-professional photography in the news

In the early aughts, academics and practitioners concerned with fake imagery were particularly focused on the implications of newly abundant photo-editing software (Greer and Gosen, 2002; Schwartz, 2003; Wheeler and Gleason, 1995). A few years later, perceptions of citizen photojournalism became a popular area of academic inquiry upon the explosion of the popularity of smartphone cameras. Several of these studies suggest that audiences appreciate many of the qualities of non-professional photography, including its authenticity and personal touch (Allan and Peters, 2015; Pantti, 2013). In some cases, the “truth value” is deemed of greater importance in citizen photography than that of professionals because of its rawness (Allan and Peters, 2015). Scholarly endeavors have begun to delve into audience perceptions of other forms of photography in the news beyond citizen photography. Mattioli and Cabitza (2024) explored credibility perceptions concerning stock photographs versus those taken by professional photojournalists, finding that participants tended to perceive stock photos as significantly less credible. Some newer studies have examined perceptions of fake imagery through the lens of AI. Americans have difficulty discerning real photos from AI-generated photos (Elon University, n.d.; Wojdynski and Shivers, 2025). Looking specifically at deepfakes, Vaccari and Chadwick (2020) found that individuals experience doubt regarding their ability to assess veracity more than being entirely deceived, and that those who express uncertainty also report lower trust in news shared on social media platforms.

Addressing audience perceptions through biometric measures

Biometric research is used to measure and analyze human characteristics and behaviors (Jain et al., 2000). These measures are useful in determining the emotional and cognitive states of individuals, including their attention, arousal, valence (emotional positivity/negativity), and memory (Kim and Kim, 2022). The face is one of the richest channels of expression, communicating both emotions and social signals (Graham and LaBar, 2012), and is the biometric indicator implemented in the present research. These emotions represent the core and universal human emotions that can be identified through facial cues. Facial expressions are temporary changes in the shape of the face due to the deformation of the face’s muscles. They serve as outward manifestations of internal emotional states, providing visible cues that reflect the individual’s arousal and valence. Facia expressions, along with textual data, can provide important cues to identify true affective states in the participants (Kim et al., 2018; Taggart et al., 2016). The facial expressions of an individual can be classified into seven fundamental emotions, which include anger, disgust, fear, happiness, neutrality, sadness, and surprise (Dores et al., 2020). Facial-expression biometric measures are also amongst the most challenging biometric measures, as human faces can be viewed from various angles with different expressions (Alrubaish and Zagrouba, 2020).

Facial action units (FAU) refer to the specific movements of individual facial muscles or muscle groups (Zhi et al., 2020). It focuses on the anatomical changes occurring within the face. Facial Action Coding System (FACS) provides a framework for identifying and categorizing these muscular movements as expressions (Clark et al., 2020). Each action corresponds to the contraction or relaxation of specific muscles, which can be objectively measured and coded according to a standardized system. Smile is a universal facial action that conveys positive emotions and has impacts on two important facial muscles which can augment facial recognition. Brow furrow is another facial action typically associated with negative emotions, such as concentration, frustration, or anger, and involves the contraction of the muscles between the eyebrows, creating vertical lines or furrows on the forehead.

Summary, Research Questions, and hypotheses

The journalistic value of truth and the photojournalistic value of the emotional portrayal of the human experience are directly at odds with the realities of dramatized, multi-use stock photography. As professional photojournalism continues to be devalued and viewed as an optional, disposable asset when budgets are tight, other forms of visuals will continue to be used to visually illustrate the happenings of the world.

Research is only beginning to reveal what this means for audiences; specifically, whether they react to the images differently. The scant research that does exist suggests that people are able to perceive the difference between false and real photojournalism, most often viewing staff-photojournalist-shot photos as more credible. As we are interested in perceptions of real and fake photos, stock photos are defined in this study as images that are created in a generic style for the purpose of multiple uses and licensing (Frosh, 2001; Machin, 2004). Previous research has also shown that negatively-valenced photographs elicit stronger audience responses than positive photographs. In light of these previous findings, we ask:

RQ1: What, if any, are the effects of photo emotions and photo source upon the self-reported arousal of viewers?

We hypothesize:

Compared to stock photos, staff photos generate (a) more positive responses (b) more negative responses, (c) more smiles, and (d) more brow furrows in participants’ facial expressions.

Compared to positive photos, negative photos generate (a) more negative responses, (b) more brow furrows in participants’ facial expressions, and (c) lower valence scores in the self-report measures.

Compared to negative photos, positive photos generate (a) more positive responses, (b) more smiles in participants’ facial expressions, and (c) higher valence scores in the self-report measures.

A positive staff photo generates (a) more positive responses and (b) more smiles in participants’ facial expressions than in other conditions (such as negative staff photos or positive stock photos).

A negative staff photo generates (a) more negative responses and (b) more brow furrows in participants’ facial expressions than in other conditions (such as positive staff photos or positive stock photos).

Method

This IRB-approved study is a mixed-factor quasi-experiment with two independent variables: valence (positive vs negative) and photo source (stock vs professional photojournalist). The research was conducted using a computer-based facial coding system called Affectiva AFFDEX in iMotions to measure human emotions. Facial expressions of participants were recorded with a C920 Pro Stream Webcam (Logitech, resolution: 1080 x 720 px, 30 frames per second) during the entire experiment. Lighting was kept constant. A video quality for every participant with a mean quality score over 95% was achieved.

Participants

Study participants were English-speaking students from a large public university enrolled in courses from a communication college in the United States. Ages ranged from 19 to 24, with a median age of 21. 97 participants completed the study, but only participants (N = 82) who recorded both facial expression recognitions and completed the survey were analyzed. The final sample size includes 41 participants for each condition. Of these, 12.2% were male (N = 10), 84.1% were female (N = 69), and 3.7% identified as other (N = 3). Additionally, 43.9% of the participants were White (N = 36), 50% identified as Black or African American (N = 41), 1.2% were Asian (N = 1), and 4.9% identified as Other Ethnicity (N = 4).

Stimuli development

Drug use-related photojournalism images depicting abuse, addiction, and recovery were selected because such content is strongly emotion-laden, making it especially appropriate for examining variations in affective valence and arousal in visual news coverage. News photos on this topic often include stock photos. Moreover, drug misuse remains highly relevant public health concerns among college-aged populations.

A non-author professional photojournalist went through search engines and award-winning photojournalism projects to find 25 potential photos taken by 12 different photographers. We selected award-winning photos because they met professional staff photojournalist standards and were pre-verified by the profession. Most of the chosen photos were captured in the past 10 years. We discussed the photos’ adherence to photojournalistic values and ultimately picked 10 stock stimuli that were highly expressive. Next, we delved into the extensive collection of stock photos available on the iStock website. Finding a suitable image that could be fairly compared to the staff photo while maintaining balance proved to be a challenge. If the two photos shared too many similarities, there would be no distinguishing factors to compare. Conversely, if too many variables differed, it would be difficult to pinpoint the cause of any discrepancies. We spent several hours scouring the collection for photos that could effectively capture the essence of the story while also featuring a similar number of people, gender, and ethnicity. Although the compositions were not always an exact match, they were captured with the same creative approach, such as employing a wide-angle lens or taking the photo from a slightly elevated position.

We conducted a pretest where 213 participants were asked to evaluate their emotional responses to three random photos from a pool of 10 stock and 10 staff photos. Valence was measured using four nine-point items on a bipolar scale: sad/happy, depressed/joyful, displeased/pleased, and distressed/delighted (Gorn et al., 2001). Arousal was measured using three nine-point items on a bipolar scale: relaxed/stimulated, calm/excited, and unaroused/aroused (Gorn et al., 2001). We selected one pair each of positive and negative photos, while both pairs had a similar level of arousal. Based on the one-way ANOVA test results, we picked one positive stock photo (M_valence = 6.30, M_arousal = 4.06) and one positive staff photo (M_valence = 5.64, M_arousal = 4.58), along with one negative stock photo (M_valence = 3.32, M_arousal = 4.48) and one negative staff photo (M_valence = 3.09, M_arousal = 4.79). Both pairs were significantly different in the valence scores but not in the arousal scores. We created four different stimuli using these four photos, each placed in the same generic local news presentation with a headline that read “Local News” and a generic subhead line.

Experimental procedure

Participants were invited to participate through an internal recruitment system at the research institution and were sent a consent letter. They were randomly assigned to view either stock photos or those taken by a professional photojournalist. Participants were given 30 seconds to view each image and their facial expression was recorded by a distant webcam. After viewing the photos, participants completed a questionnaire about their emotional responses to each photo and demographic information. Participants were compensated with a $10 gift card.

Dependent measures

Facial expression analysis

Video cameras captured detailed close-ups of participants’ faces for facial expression analysis. The Affectiva software, utilizing a frame-by-frame analysis, categorizes images and videos of facial expressions according to the emotions displayed. The efficacy and reliability of the Affectiva AFFDEX software have been validated through studies on static images (Stöckli et al., 2018) and videos (Taggart et al., 2016). These recordings underwent digital coding by Affectiva Affdex within the iMotions platform (version 8.3). The study targeted four measures to evaluate the proposed hypotheses: the counts of positive and negative responses, and images depicting smiling and brow furrowing. Each participant’s facial expression was assessed on a probability scale ranging from 0 to 100%. A response that surpasses the 50% probability threshold was recorded as a specific facial expression occurrence.

AdSAM (Valence and arousal)

AdSAM®, an adaptation of Lang et al. (1993) Self-Assessment Manikin, was used to gauge viewers’ emotional reactions to media stimuli along three affective dimensions—valence, arousal, and dominance (Morris 1995)—using three parallel rows of stylized “manikin” figures that visually depict each state. After view the images, respondents selected the manikins that best matched their experience (Figure 1).

Figure 1.

Staff photo, negative tone.

Data analysis models

When analyzing facial expressions, it is common to have instances where participants do not exhibit any specific facial expressions or neutral expressions, resulting in an excess of zeros. This research used zero-inflated Poisson regression to investigate the correlation between four facial expression variables (positive/negative responses, smile, and brow furrow) and two photo sources. Zero-inflated Poisson regression is a statistical technique that is useful in handling count data with a high number of zero counts. Additionally, a two-way Univariate Analysis of Variance (ANOVA) was employed to analyze the effects of photo source (staff vs stock) and emotion (positive vs negative) on the continuous self-report measures through Attitude Self-Assessment Manikin (AdSAM) (i.e., valence and arousal) (Figure 2).

Figure 2.

Stock photo, negative tone.

Results

Staff photos generally produced greater effects than stock photos. While the sources of the photos did not result in significantly different arousal scores, some nuanced interaction effects were observed. Notably, negative staff images elicited significantly stronger arousal levels than negative stock photos. Compared to stock photos, staff photos generated more reactions, including more smiles, and more brow furrows overall. Specifically, positive staff photos received more positive responses and more smiles, and negative staff photos generated more negative responses and negative brow furrows when compared to positive and negative stock photos, respectively. Further, when compared to positive photos, negative photos generated more negative responses, more brow furrow in respondents’ facial expressions, and lower valence score in the self-report measure, and compared to negative photos, positive photos generated more positive responses, more smiles in respondents’ facial expressions, and higher valence score in the self-report measures (Figure 3).

Figure 3.

Staff photo, positive tone.

Research Question 1 asked about the effects of photo source and emotion upon the arousal score in the self-report measure. The main effect of photo source (M_staff = 4.46, SD_staff = 1.9; M_stock = 4.30, SD_stock = 2.39) did not generate varied self-report arousal scores (F (1, 160) = 0.259, p = 0.611). However, there was a significant interaction effect between photo emotion and photo source on self-reported arousal levels (F (1,160) = 5.710, p = 0.018). Specifically, the negative staff image (M = 4.07, SD = 1.84) triggered stronger arousal than the negative stock image (M = 3.17, SD = 1.95, p < 0.042). In contrast, both the positive staff image (M = 4.85, SD = 1.91) and positive stock image (M = 5.44, SD = 2.26) elicited similar reactions (p = 0.186). Additionally, if compared within the two staff images, both negative staff (M = 4.07, SD = 1.84), and positive staff photos (M = 4.85, SD = 1.91) generated similar levels of arousal (p = 0.078). In contrast, if compared within the two stock images, the positive stock photo (M = 5.44, SD = 2.26) generated significantly stronger arousal than the negative stock photos (M = 3.17, SD = 1.95, p < 0.001). The main effect of emotions (positive or negative) was significant on arousal scores, as well. Positive images (M = 5.15, SD = 2.10) generated a higher level of arousal than negative ones (M = 3.62, SD = 1.94), F (1, 160) = −23.976, p < .01 (Figure 4).

Figure 4.

Stock photo, positive tone.

H1 hypothesized that staff photos would generate more (a) positive responses, more (b) negative responses, (c) more smiles, and (d) more brow furrows in participants’ facial expressions compared with the stock photos. The result shows that there was a statistically significant difference in more positive responses generated by staff photos (β = 1.524, SE = 0.076, p < .01), more smiles (β = 1.186, SE = 0.066, p < .01) and more brow furrow (β = 1.953, SE = 0.091, p < .01) in participants’ facial expressions compared to stock photos, which provide support for Hypotheses 1(a), (c), and (d). There was no significant difference between staff and stock photos in generating negative responses (β = −0.068, SE = 0.081, p = 0.397).

Hypothesis 2 proposed that, overall, negative photos generate (a) more negative responses, (b) more brow furrow in respondents’ facial expressions, and (c) a lower valence score in the self-report measure than positive photos. All three parts of Hypothesis 2 were supported, with a statistically significant difference in the number of negative responses (β = −0.602, SE = 0.081, p < .01) the number of brow furrows (β = −1.805, SE = 0.091, p < .01) generated by participants in response to negative photos, and the self-report measure indicated a lower valence score (F (1,160) = 680.75, p < .01) for negative photos compared to positive photos.

Hypothesis 3 proposed that positive photos would generate (a) more positive responses, (b) more smiles in respondents’ facial expressions, and (c) a higher valence score in the self-report measure than negative photos. The results of the study showed a statistically significant difference in the number of positive responses (β = 0.685, SE = 0.076, p < .01) and number of smiles (β = 0.886, SE = 0.066, p < .01) generated by participants in response to positive photos. Moreover, the self-report measure indicated a higher valence score (F (1,160) = 680.75) for positive photos compared to negative photos. This finding supports all three parts of Hypothesis 1. Thus, Hypotheses 3 (a), (b), and (c) were all supported.

In Hypothesis 4, it was hypothesized that the presence of a positive staff photo would generate (a) more positive responses and (b) more smiles in participants’ facial expressions than in other conditions, such as negative staff photos or positive stock photos. The results showed that there was a statistically significant increase in positive responses (β = 0.146, SE = 0.038, p < .01) and more smiles (β = 0.171, SE = 0.033, p < .01) when a positive and staff photo was present compared to other conditions. Therefore, both parts of H4 are supported.

Hypothesis 5 asserted that the presence of a negative staff photo would generate (a) more negative responses and (b) more brow furrows in participants’ facial expressions than in other conditions, such as positive staff photos or positive stock photos. The results support this hypothesis, as that the negative staff photo generates more negative responses (β = −0.132, SE = 0.040, p = .01) and more brow furrow (β = 0.344, SE = 0.045, p < .01) compared to other conditions.

Discussion and conclusions

The photojournalistic mission of presenting truthful depictions of the emotional human experience conflicts with the prevalence of false photography online. This phenomenon underscores a broader trend wherein professional photojournalism faces diminishing valuation and is considered dispensable. Consequently, alternative visuals such as stock photos and AI are utilized, which are directly at odds with professional photo values in that they are neither truthful nor do they depict true emotions. As viewers become more uncertain about the veracity of news visuals, trust in the news industry may suffer (Vaccari and Chadwick, 2020).

Staff photos generated stronger emotional responses compared to stock images, suggesting genuine emotion is detectable and valued. Staff photos generated more positive frames, more smiles, and more brow furrows in participants’ facial expressions compared with the stock photos. This held true for both positive photos and negative photos. This result adds to a previous study have suggested a preference for professional staff photographs in the news in credibility perceptions (Mattioli and Cabitza, 2024). This distinction also mirrors Tucker and Riggio’s (1988) observation that spontaneous expressions bypass skepticism by aligning with viewers’ intuitive expectations of authenticity. Further, as Friend and Singer (2015) noted, sincerity in communication arises not from technical perfection but from the seamless integration of emotional and contextual cues—a dynamic exemplified in staff photos’ ability to evoke visceral reactions. Further, respondents who were exposed to negative emotional photographs from staff resources had an increase in levels of arousal compared to those who did not, hinting at a greater impact of the genuine emotion depicted in professional, staff-taken photography. These results align with research that suggests the valuation of negative photography in the news and that negative images elicit stronger responses (Eastwood et al., 2001; Fenske and Eastwood, 2003; Fox et al., 2001; Calvo and Lang, 2004; Gelpi and Gartner, 2011; Knobloch et al., 2003; Rössler et al., 2011).

These responses suggest that audiences intuitively interpret raw, unfiltered emotion as a marker of authenticity. Emotional engagement amplifies perceptions of truthfulness (Barrett et al., 2007), and parallels findings from Tucker and Riggio (1988), who demonstrated that spontaneous emotional expressions (analogous to staff photos) are perceived as more genuine than posed ones (analogous to stock images), due to their alignment with unscripted social contexts. The conceptualization of “truth” in this study extends beyond factual accuracy to encompass perceived authenticity—the extent to which a photo is seen as an unmanipulated, human-captured representation of reality. Staff photos, rooted in photojournalistic ethics (Lough, 2021; Silva and Eldridge II, 2020), signal authenticity through raw emotion. Consistency between emotional content (a staff photo’s genuine sadness) and contextual framing (a story on addiction) enhances perceived sincerity (Friend and Singer, 2015).

The implications of these findings extend beyond drug-related imagery. Authentic visuals are not merely aesthetic choices but ethical imperatives in sustaining public engagement. As Tucker and Riggio (1988) noted, spontaneous emotional expressions (e.g., raw depictions of suffering in conflict zones) resonate more deeply than staged alternatives. In an era of AI, deepfakes, and synthetic media, institutions prioritizing human-captured visuals may better inoculate audiences against disinformation, which often weaponizes artificial emotional appeals or may hallucinate even in benign contexts.

The nature of studying real photographs means that single variables are not easily isolated. While every precaution was taken to find suitably comparable photographs, it is possible that differences between the paired photographs other than emotion and source could have resulted in the statistical differences in perceptions. Moreover, although drug use imagery provides a theoretically appropriate context for studying emotional valence, responses to other issue domains may differ. Emotional reactions to topics such as political violence, climate change, or disaster preparation may follow distinct affective patterns. Future research should examine whether the observed effects generalize across a broader range of news topics and visual contexts. In addition, the respondent sample consisted of college students, which may limit the generalizability of the findings. Finally, the use of Affectiva in this study warrants caveats. First, while the participant sample was racially diverse, the present study cannot fully evaluate potential differential performance of facial expression recognition systems across facial phenotypes associated with race, skin tone, and gender. Prior research suggests that such systems may exhibit uneven accuracy due to variation in training data representation and culturally patterned expression norms (Mattioli and Cabitza, 2024). In addition, the gender composition of the image stimuli differed across valence conditions. Second, viewing static and video stimuli in a laboratory differs meaningfully from everyday, incidental exposure. Controlled viewing minimizes head movement and distractors, but it can also heighten self-consciousness and attenuate spontaneous affective display, which may partly account for the modest effect sizes observed. Finally, Affectiva currently codes a limited subset of action units and blends these into a small set of categorical emotions; nuanced or blended states are therefore under-detected.

This study demonstrates that authenticity is a currency of trust in visual communication. There is continued merit of real, truthful news photographs taken by on-staff photojournalists. Staff-taken photos elicited stronger emotional and physiological responses than stock images, not only because they are “professionally-made,” but because they embody photojournalism’s core mission: to reflect unfiltered human experience. Despite the thinning of photojournalism staffs and increased technologies and availabilities for visuals there remain a deep valuation of professional photojournalistic images which depict genuine emotions. Without knowing the source of the photo, viewers had stronger negative and positive reactions to these photographs. In several ways, the results of this study are positive for the photojournalism industry, while concerning for the state of visual news and visual misinformation.

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Tara Marie Mortensen

Author biographies

Tara Marie Mortensen is an Associate Professor in the School of Journalism and Mass Communications at the College of Information and Communications at the University of South Carolina. She teaches graphics-related classes and researches the evolving visual media landscape.

Carrie Jingyi Xiao is a doctoral student in the School of Journalism and Mass Communications at the University of South Carolina, Columbia, South Carolina, USA.

Taylor Jing Wen is an Associate Professor in the School of Journalism and Mass Communications, College of Information and Communications at the University of South Carolina, Columbia, South Carolina, USA.

References

Aiello

(2016) Taking stock. Ethnography Matters. https://ethnographymatters.net/blog/2016/04/28/taking-stock

Allan

Peters

(2015) The “public eye” or “disaster tourists.”. Digital Journalism 3(4): 477–494. https://doi.org/10.1080/21670811.2015.1034517

Alrubaish

Zagrouba

(2020) The effects of facial expressions on face biometric system’s reliability. Information 11(10): 485. https://doi.org/10.3390/info11100485

Arthur

(2021) Phone Photography for Everybody: Iphone Photojournalism Techniques. Amherst Media.

Barnham

(2023) Photojournalism, what’s happened? LinkedIn. https://www.linkedin.com/pulse/photojournalism-whats-happened-andy-barnham/

Barrett

Mesquita

Ochsner

, et al. (2007) The experience of emotion. Annual Review of Psychology 58: 373–403. https://doi.org/10.1146/annurev.psych.58.110405.085709

Berger

Milkman

(2012) What makes online content viral? Journal of Marketing Research 49(2): 192–205. https://doi.org/10.1509/jmr.10.0353

Bock

(2023) Visual media literacy and ethics: images as affordances in the digital public sphere. First Monday 28(7): https://doi.org/10.5210/fm.v28i7.13233

Bradley

Lang

(1994) Measuring emotion: the self-assessment manikin and the semantic differential. Journal of Behavior Therapy and Experimental Psychiatry 25(1): 49–59. https://doi.org/10.1016/0005-7916(94)90063-9

10.

Bransford

Johnson

(1972) Contextual prerequisites for understanding: some investigations of comprehension and recall. Journal of Verbal Learning and Verbal Behavior 11(6): 717–726. https://doi.org/10.1016/S0022-5371(72)80006-9

11.

Brennen

(1998) Strategic competition and the value of photographers’ work: photojournalism in. American Journalism 15(2): 59–77. https://doi.org/10.1080/08821127.1998.10731971

12.

Calvo

Lang

(2004) Gaze patterns when looking at emotional pictures: motivationally biased attention. Motivation and Emotion 28(3): 221–243. https://doi.org/10.1023/B:MOEM.0000040153.26156

13.

Clark

Kessinger

Duncan

, et al. (2020) The facial action coding system for characterization of human affective response to consumer product-based stimuli: a systematic review. Frontiers in Psychology 11: 920. https://doi.org/10.3389/fpsyg.2020.00920

14.

Costa

(1950) The Complete Book of Press Photography. The National Press Photographer’s Association.

15.

Dores

Barbosa

Queirós

, et al. (2020) Recognizing emotions through facial expressions: a largescale experimental study. International Journal of Environmental Research and Public Health 17(20): 7420. https://doi.org/10.3390/ijerph17207420

16.

Eastwood

Smilek

Merikle

(2001) Differential attentional guidance by unattended faces expressing positive and negative emotion. Perception & psychophysics 63(6): 1004–1013.

17.

Elliott

(2011) Ethical responsibilities and the power of pictures. In: Lester

(ed) Images that Injure: Pictorial Stereotypes in the Media. Praeger, pp. 7–14.

18.

Elon University (n.d.) . AI and Politics 2024. Imagining the Digital Future. Available at https://imaginingthedigitalfuture.org/reports-and-publications/ai-politics-24/ (accessed on February 21, 2025).

19.

Emmett

(2011) Haunting Photos: Differential Perception and Emotional Response to the Archetypes of News Photography: A Study of Visual Reception Factored by Gender and Expertise. College Park: University of Maryland.

20.

Fahmy

Bock

Wanta

(2014) Visual Communication Theory and Research: A Mass Communication Perspective. Springer.

21.

Fenske

Eastwood

(2003) Modulation of focused attention by faces expressing emotion: evidence from flanker tasks. Emotion 3(4): 327–343. https://doi.org/10.1037/1528-3542.3.4.327

22.

Fenske

Raymond

Kunar

(2004) The affective consequences of visual attention in preview search. Psychonomic Bulletin & Review 11(6): 1055–1061. https://doi.org/10.3758/BF03196736

23.

Friedman

(1979) Framing pictures: the role of knowledge in automatized encoding and memory for gist. Journal of experimental psychology: General 108(3): 316.

24.

Friend

Singer

(2015) Online Journalism Ethics: Traditions and Transitions. Routledge.

25.

Frosh

(2001) Inside the image factory: stock photography and cultural production. Media, Culture & Society 23(5): 625–646.

26.

Gade

(2011) Postmodernism, uncertainty, and journalism. In: Lowrey

Gade

(eds) Changing the News: The Forces Shaping Journalism in Uncertain Times. Routledge, pp. 63–82.

27.

Gelpi

Gartner

(2011) Affect and photos of success and failure in war. In: Paper Presented at American Political Science Association Annual Meeting.

28.

Godulla

Seibert

Planer

(2021) Whose pictures, whose reality? Lines of tradition in the development of topics, negativity, and power in the photojournalistic competition world press photo. Journalism and Media 2(4): 758–807. https://doi.org/10.3390/journalmedia2040045

29.

Gorn

Tuan Pham

Yatming Sin

(2001) When arousal influences ad evaluation and valence does not (and vice versa). Journal of Consumer Psychology 11(1): 43–55. https://doi.org/10.1207/S15327663JCP1101_4

30.

Graham

LaBar

(2012) Neurocognitive mechanisms of gaze-expression interactions in face processing and social attention. Neuropsychologia 50(5): 553–566. https://doi.org/10.1016/j.neuropsychologia.2012.01.019

31.

Greenwood

Thomas

(2017) Locating the journalism in citizen photojournalism: The use and content of citizen-generated imagery. Photojournalism and citizen journalism. Routledge, 298–316.

32.

Greer

Gosen

(2002) How much is too much? Assessing levels of digital alteration of factors in public perception of news media credibility. Visual Communication Quarterly 9(3): 4–13. https://doi.org/10.1080/15551390209363485

33.

Horton

(1989) The Picture: An Associated Press Guide to Good News Photography. The Associated Press.

34.

Jain

Hong

Pankanti

(2000) Biometric identification. Communications of the ACM 43(2): 90–98. https://doi.org/10.1145/328236.328110

35.

Kim

(2022) Quantifying emotions in architectural environments using biometrics. Applied Sciences 12(19): 9998. https://doi.org/10.3390/app12199998

36.

Kim

, et al. (2018) Building emotional machines: recognizing image emotions through deep neural networks. IEEE Transactions on Multimedia 20(11): 2980–2992. https://doi.org/10.48550/arXiv.1705.07543

37.

Klein-Avraham

Reich

(2016) Out of the frame: a longitudinal perspective on digitization and professional photojournalism. New Media & Society 18(3): 429–446. https://doi.org/10.1177/1461444814545289

38.

Knobloch

Hastall

Zillmann

Callison

(2003) Imagery effects on the selective reading of Internet newsmagazines. Communication Research 30(1): 3–29.

39.

Kobré

(2011) Photojournalism: The Professionals’ Approach. Routledge.

40.

Kovach

Rosenstiel

(2021) The Elements of Journalism: What Newspeople Should Know and the Public Should Expect. 4th rev. Crown Publishing Group.

41.

Lang

Greenwald

Bradley

, et al. (1993) Looking at pictures: affective, facial, visceral, and behavioral reactions. Psychophysiology 30(3): 261–273. https://doi.org/10.1111/j.1469-8986.1993.tb03352.x

42.

Lester

(2018) Visual Ethics: A Guide for Photographers, Journalists, and Filmmakers. Taylor & Francis.

43.

Lester

Martin

Smith-Rodden

(2022) Visual Ethics: A Guide for Photographers, Journalists, and Media Makers. Routledge.

44.

Lough

(2021) Judging photojournalism: the metajournalistic discourse of judges at the best of photojournalism and pictures of the year contests. Journalism Studies 22(3): 305–321. https://doi.org/10.1080/1461670X.2020.1867001

45.

Machin

(2004) Building the world’s visual language: The increasing global importance of image banks in corporate media. Visual communication 3(3): 316–336.

46.

Martin-Kratzer

(2005) The Emotional and Cognitive Processing of Negative News Photographs. Doctoral dissertation, University of Missouri--Columbia.

47.

Mattioli

Cabitza

(2024) Not in my face: challenges and ethical considerations in automatic face emotion recognition technology. Machine Learning and Knowledge Extraction 6(4): 2201–2231. https://doi.org/10.3390/make6040109

48.

Merrill

(1974) The Imperative of Freedom: A Philosophy of Journalistic Autonomy. Freedom House.

49.

Moran

Shaikh

(2022) Robots in the news and newsrooms: unpacking meta-journalistic discourse on the use of artificial intelligence in journalism. Digital Journalism 10(10): 1756–1774. https://doi.org/10.1080/21670811.2022.2085129

50.

Morris

(1995) Observations: SAM: the self-assessment manikin; an efficient cross-cultural measurement of emotional response. Journal of Advertising Research 35(6): 63–68. https://doi.org/10.1080/00218499.1995.12466497

51.

Mortensen

Gade

(2018) Does photojournalism matter? News image content and presentation in the Middletown (NY) Times Herald-Record before and after layoffs of the photojournalism staff. Journalism & Mass Communication Quarterly 95(4): 990–1010.

52.

Murphy

Flynn

(2022) Deepfake false memories. In: Nash

Newman

(eds) Memory Online. Routledge, pp. 112–124.

53.

Nash

(2018) Changing beliefs about past public events with believable and unbelievable doctored photographs. Memory 26(4): 439–450. https://doi.org/10.1080/09658211.2017.1364393

54.

Newman

Garry

Unkelbach

, et al. (2015) Truthiness and falsiness of trivia claims depend on judgmental contexts. Journal of Experimental Psychology: Learning, Memory, and Cognition 41(5): 1337–1348. https://doi.org/10.1037/xlm0000099

55.

Newton

(1998) The burden of visual truth: the role of photojournalism in mediating reality. Visual Communication Quarterly 5(4): 4–9. https://doi.org/10.1080/15551399809363390

56.

Nilsson

(2021) Expendable or valuable? Photojournalism in five Swedish newspapers affected by organizational changes. Journalistica 15(1): 33–58.

57.

Pantti

(2013) Getting closer? Encounters of the national media with global images. Journalism Studies 14(2): 201–218. https://doi.org/10.1080/1461670X.2012.718551

58.

Porlezza

Schapals

(2024) AI ethics in journalism (studies): an evolving field between research and practice. Emergency Medicine 2(3): 356–370. https://doi.org/10.1177/27523543241288818

59.

Rössler

Bomhoff

Haschke

, et al. (2011) Selection and impact of press photography. An empirical study on the basis of photo news factors. Communications 36(4): 415–439. https://doi.org/10.1515/comm.2011.021

60.

Schwartz

(2003) Professional oversight: policing the credibility of photojournalism. In: Image Ethics in the Digital Age, 27–51.

61.

Silva

MFS

Eldridge II

(2020) The Ethics of Photojournalism in the Digital Age. Routledge.

62.

Singletary

Lamb

(1984) News values in award-winning photos. Journalism Quarterly 61(1): 104–233. https://doi.org/10.1177/107769908406100114

63.

Skinner

(2010) Sports Photography: How to Capture Action and Emotion. Skyhorse Publishing, Inc.

64.

Sonderman

(2013) Use of Generic Photos Can Be Dangerous for Illustrating News Stories. Poynter. https://www.poynter.org/reporting-editing/2013/use-of-generic-photos-to-illustrate-news-stories-can-be-dangerous/

65.

Stöckli

Schulte-Mecklenbeck

Borer

, et al. (2018) Facial expression analysis with AFFDEX and FACET: a validation study. Behavior Research Methods 50: 1446–1460. https://doi.org/10.3758/s13428-017-0996-1

66.

Taggart

Dressler

Kumar

, et al. (2016) Determining emotions via facial expression analysis software. In: Proceedings of Student-Faculty Research Day, CSIS. Pace University.

67.

Thurlow

Aiello

Portmann

(2020) Visualizing teens and technology: A social semiotic analysis of stock photography and news media imagery. New Media & Society 22(3): 528–549.

68.

Tucker

Riggio

(1988) The role of social skills in encoding posed and spontaneous facial expressions. Journal of Nonverbal Behavior 12(2): 87–97.

69.

Vaccari

Chadwick

(2020) Deepfakes and disinformation: exploring the impact of synthetic political video on deception, uncertainty, and trust in news. Social Media+Society 6(1): 2056305120903408. https://doi.org/10.1177/2056305120903408

70.

Wallon-Hárs (2023) How to sell stock photos on alamy: wahaviblog about stock photography. WahaviBlog. https://wahaviblog.com/en/sell-stock-photos-alamy

71.

Wheeler

Gleason

(1995) Photography or photofiction: an ethical protocol for the digital age. Visual Communication Quarterly 2(1): 8–12. https://doi.org/10.1080/15551393.1995.10387515

72.

Williams

Newton

(2009) Visual Communication: Integrating Media, Art, and Science. Routledge.

73.

Wojdynski

Shivers

(2025, March 13 – 15). Challenging perceptions: AI-generated images and visual credibility. Paper presented at the annual meeting of the Association for Education in Journalism and Mass Communications (AEJMC), San Francisco, CA, United States.

74.

Zelizer

(1995) Journalism’s “last” stand: wirephoto and the discourse of resistance. Journal of Communication 45(2): 78–92. https://doi.org/10.1111/j.1460-2466.1995.tb00729.x

75.

Zhi

Liu

Zhang

(2020) A comprehensive survey on automatic facial action unit analysis. The Visual Computer 36(5): 1067–1093. https://doi.org/10.1007/s00371-019-01707-5