Objective The emotional impact of the tobacco-warning images proposed by the European Commission to reduce tobacco consumption is evaluated in the context of the International Affective Picture System, a well-established procedure for investigating appetitive (approach) and defensive (avoidance) motivational tendencies evoked by images.
Methods In a cross-sectional study, 597 healthy male and female volunteers (from the University of Granada, the University of Balearic Islands and four different schools of Valencia and Balearic Islands) distributed in six age groups (13–14, 15–16, 17–18, 19–20, 21–22 and over 23 years old) and four smoking status groups (non-smokers, one-time smokers, occasional smokers and heavy smokers) rated their emotional responses to 35 European tobacco-warning images together with 42 pleasant and 42 unpleasant International Affective Picture System pictures using the valence and arousal scales of the Self-Assessment Manikin.
Findings The results of the study indicate that the majority of the tobacco-warning images (83%) were distributed within the unpleasant space and ranged from moderately unpleasant to very unpleasant. However, a small but significant number of images (17%) were also distributed within the pleasant space, ranging from moderately pleasant to very pleasant. Only four unpleasant pictures were rated as highly arousing (11.4%). Women, the older age groups (over 17 years old), and occasional smokers evaluated these images as significantly more arousing than the other groups.
Conclusion Findings suggest that the capability of the European tobacco-warning images to prompt negative attitudes to reduce tobacco consumption might not extend to the general population but would be limited to specific target groups.
- Packaging and labelling
- warning images
- avoidance behaviour
- defensive motivation
- global health
- smoking-caused disease
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- Packaging and labelling
- warning images
- avoidance behaviour
- defensive motivation
- global health
- smoking-caused disease
Tobacco consumption is the single greatest cause of avoidable death in Europe.1 It has been estimated that 25% of all cancer deaths and 15% of all deaths in the European Union (EU) can be attributed to smoking and that over 13 million EU citizens suffer from a serious chronic disease as a result of their smoking.1 Since the 1980s, there have been various legislative initiatives to curb tobacco use among citizens. The aim has been to present certain restrictions in the marketing of tobacco and to regulate the labelling of tobacco products. Sentences such as ‘Smoking can be harmful to your health’ and other warning texts have been a common method of informing the public about the risks of tobacco.2–4
Although various studies have shown that warning texts on cigarette packages have a positive influence on the reduction of smoking,5 ,6 it has been argued that warnings consisting purely of text may not sufficiently capture an individual's attention to allow the message to be processed.7–9 As a consequence, countries, such as Canada, Brazil and Australia, have introduced pictures showing the unhealthy effects of smoking.4 ,10 In 2003, the European Commission developed a series of pictorial warnings to be used on tobacco packages. These warnings show the negative impact of tobacco by presenting powerful visual images.11 Pictorial warnings are currently not mandatory in the EU. However, the Commission is encouraging their wider use. Countries such as the UK, Belgium and Romania have already adopted these measures, whereas others, such as Spain, have done it recently.
Unpleasant pictorial warnings can be a successful tool to prevent people from starting to smoke and induce consumers to quit smoking.7 ,12 It has been shown, for instance, that the negative emotional reactions generated by warning images are associated with dispositions to give up smoking habits, to reduce the amount of cigarettes smoked and to develop negative attitudes towards smoking.13 ,14
The impact of image to influence attitudes is well known by tobacco industry. Cigarette packaging is among the most prominent forms of tobacco marketing, communicating calm, wealth, elegance, optimism and independence.15 Tobacco industry tries to influence people attitudes, especially teens, by making them to see tobacco products attractive and desired. Neuroscience research on emotion has consistently demonstrated that visual stimuli affect attitude and attention, producing measurable and reliable changes in physiology.16–18 According to Lang's Bio-informational Model,16 ,19 emotions are dispositions to action and reflect the motivational system that is engaged. Emotional visual stimuli can prompt states of highly focused attention reflecting central activation and preparation for action.20–22 A key assumption in Lang's model is that the neural circuits underlying emotion have direct connections to the brain's primary motivational systems: the appetitive–approach system, associated with pleasant emotions, and the defensive–avoidance system, associated with unpleasant emotions.23 Extensive research on these two motivational systems has been conducted using the International Affective Picture System (IAPS), a catalogue of hundreds of pictures with normative data on their subjective valence (pleasant–unpleasant) and arousal (low–high intensity).24 It has been consistently demonstrated that the distribution of the pictures along these two dimensions forms a boomerang-like pattern with two arms that extend from a neutral affective zone with minimum intensity to the two extremes of appetite and unpleasant maximum intensity. This organisation appears to be wholly compatible with a two-dimensional structure of the motivational system: an appetitive–approach (upper arm) and a defensive–avoidance (lower arm) motivational system that varies along an activation dimension.16 ,25
The impact of tobacco-warning images and their potential effect for evoking negative attitudes towards tobacco consumption can be evaluated in the context of these two motivational systems. If tobacco-warning images are very unpleasant and highly arousing, they might promote focused attention and negative attitudes, thereby facilitating activation of the defensive motivational system and avoidance behaviours. The present study examines the subjective emotional impact of the tobacco-warning pictures employed in the European campaign and compares them with the impact of standard pleasant, neutral and unpleasant IAPS pictures. Furthermore, the present study also investigates the possible variation of this impact with respect to gender, age and smoking status. We suggest that important insights and suggestions concerning the potential effectiveness of the European campaign can be produced by this evaluation procedure, and it could generalise to others pictographic campaigns.
Five hundred and ninety-seven healthy subjects (369 women and 228 men) between 13 and 44 years old participated in the study. Volunteers were recruited from the University of the Balearic Islands (N=92), the University of Granada (N=186) and four Spanish schools of Valencia (N=219) and Balearic Islands (N=100). Table 1 summarises the characteristics of the participants concerning gender, age and smoking status. The experimenters interviewed the head of each academic institution to present the study and to request authorisation. The local ethics committee of each institution approved the experimental protocol.
Eighty-four images (42 pleasant and 42 unpleasant) were selected from the Spanish version of the IAPS (pleasant: 8191, 8186, 8185, 8178, 8161, 8021, 8496, 8400, 8341, 8193, 7340, 7330, 5811, 5779, 5551, 4614, 4610, 2395, 2346, 2341, 2340, 2311, 2224, 2071, 2058, 1610, 1463, 1419; neutral: 2840, 2520, 2514, 2487, 2485, 2480, 2383, 2372, 2191, 2190, 7025, 7010, 7009, 7004, 7002, 7235, 7233, 7224, 7207, 7175, 7950, 7205, 7187, 7186, 5395, 2575, 2515, 2210; unpleasant: 9621, 9620, 9584, 9582, 9495, 9471, 9470, 9342, 9341, 6910, 6900, 5972, 5971, 3022, 2900, 2800, 2276, 2130, 2120, 2095, 1525, 1300, 1274, 1270, 1220, 1205, 1120, 1050). Selection was based on the following criteria: pictures with valence scores between 1 and 4.99 were classified as being from very unpleasant to lowly unpleasant and between 5 and 9 as being from lowly pleasant to very pleasant. Because teenagers were involved in this study, erotica, mutilated bodies and violent pictures were not included in the selected set of images. The pleasant images included sports, food, nature scenes, babies and happy families. Unpleasant images included pollution, disgusting animals and angry faces. Neutral (ie, lowly pleasant and unpleasant) images included non-emotional faces, buildings and utensils. This set of pictures was also balanced on the dimension of arousal: pictures with arousal scores between 6 and 9 were classified as strongly arousing, between 6 and 4 as moderately arousing and between 4 and 1 as weakly arousing.
The tobacco-warning images used in the study were the 35 images designed to be printed on cigarette packets as part of the tobacco control campaigns adopted by the European Commission (http://ec.europa.eu/health/ph_determinants/life_style/Tobacco/Documents/com_1452_a_en.pdf). Tobacco warnings without images were excluded. To avoid an explicit association of the images with the campaign, any printed words were excluded. Their graphical layouts were therefore indistinguishable from those of the IAPS pictures.
The complete set of images was divided into two experimental sets of 60 and 59 images with 18 and 17 tobacco campaign images in each set, respectively (to make equal the number of images in each set (60), one image used in the Brazilian tobacco campaign was included in the 59 images set with similar features to 35 h image, this image was not included in the statistical analyses). In order to avoid influences from presentation order, each set was presented in three fixed semi-randomised orders, which changed the position of a particular image within the entire set of images and avoided that images of the same category would be presented consecutively more than twice.
The study was conducted during December 2008 to April 2010. The experimental session was conducted in a dimly lit room with desks placed in rows in front of a slide projection screen. Maximum size of the projected image was 2.20×2.50 m. The desks were arranged to allow the screen to be perfectly visible to each participant. Experimental sessions were held in wide classes (between 50 and 100 desks), although no >27 participants took the test at a time. Each participant was at least 3.5 m away from the projected image, and the male/female ratio was no >1:2 (or 2:1) for any single group session. A computer-projector multimedia system controlled the instructions and stimuli presentation to minimise differences between sessions.
Before the start of the experimental session, each participant provided written informed consent. The participants were told that their task was to evaluate emotional pictures. The relationship between the pictures and the tobacco-warning images was not made explicit to the participants. The participants were given instructions to assess each image using the Self-Assessment Manikin scales (SAM; see below). Finally, the participants practiced the assessment procedure with three control images.
The rating of each image began with a preparation slide (‘Get ready to rate the next image’) that lasted for 3 s, followed by a 6 s picture observation period. During the next 15 s interval, participants were asked to rate the feelings that they experienced while viewing the picture along the dimensions of hedonic valence and emotional arousal using the paper-and-pencil version of the SAM scales.26 Each SAM scale is represented by five humanoid figures (and four intervals between figures) ranging from a frowning to a smiling face (for the valence dimension) and from a sleepy looking to an apparently agitated figure (for the arousal dimension). The ratings were converted to numbers ranging from 1 (lower extreme) to 9 (upper extreme). The participants rated a total of 60 images during the experimental session. After completion of the task, the participants completed a questionnaire on smoking habits.
The mean and SD of the responses were calculated for each picture on the two SAM scales. Next, all picture scores were represented in the two-dimensional valence–arousal space. Tobacco-warning pictures with scores >5 on the 1–9 valence scale were considered to be within the pleasant cluster, whereas pictures with scores <5 were considered to be within the unpleasant cluster. This strategy is common in the IAPS research27 and was adopted to compare the present distribution of the tobacco-warning images with previous reports.28 ,29 Then, Pearson correlations between valence and arousal, together with a regression line using arousal as the independent variable and valence as the dependent variable, were obtained for the tobacco-warning pictures situated within the unpleasant cluster. Due to the small number of images, this analysis was not performed for the tobacco images situated within the pleasant cluster. The same procedure was used to examine differences due to gender, age and smoking status. Differences in the slope of the regression lines between subgroups were analysed using the Potthoff method.30 Finally, in order to study the influence of gender, age and smoking status on tobacco-warning evaluations, two multiple regression analyses were performed using as dependent variable the mean valence score and the mean arousal score of unpleasant tobacco-warning images, respectively.
Valence and arousal ratings of pleasant, neutral, unpleasant and tobacco-warning images
The average ratings of the IAPS images were very similar to those obtained for the Spanish reference norms28 ,29 for pleasant (valence: mean=7.33, SD=0.48; arousal: mean=5.34, SD=1.29), neutral (valence: mean=5.21, SD=0.45; arousal: mean=3.30, SD=0.39) and unpleasant images (valence: mean=3.00, SD=0.81; arousal: mean=4.90, SD=0.57). The average ratings of the tobacco-warning images ranged from 1.82 to 7.36 (mean=3.76, SD=0.73) on the valence scale and from 3.18 to 6.75 (mean=4.93, SD=1.45) on the arousal scale (see table S1 in the supplementary material).
Figure 1 represents the distribution of the tobacco-warning and IAPS images in the two-dimensional space formed by valence (y axis) and arousal (x axis). Of the 35 tobacco-warning images, 29 are situated within the unpleasant cluster (lower arm of the boomerang), implying that they were assessed as being from moderately negative to very negative. Six tobacco-warning images are situated within the pleasant cluster (upper arm of the boomerang), implying that they were assessed as being from moderately positive to very positive. For the unpleasant tobacco-warning images, a significant negative correlation was found between the ratings of valence and arousal (r=−0.815, p<0.0001). A linear regression analysis using arousal as the independent variable and valence as the dependent variable revealed a significant negative slope (R2=0.664, t=−7.30, p<0.0001).
Gender-related differences in the ratings of tobacco-warning images
Figure 2 shows the distributions of the 35 tobacco-warning images in the valence–arousal space as a function of gender. Both men and women evaluated six images as pleasant (rating >5) and 29 as unpleasant (rating <5). For the unpleasant tobacco-warning images, both women and men showed a significant negative correlation between the ratings of valence and arousal (men: r=−0.718, p<0.0001; women: r=−0.831, p<0.0001) and a significant negative linear regression slope (men: R2=0.516, t=5.36, p<0.0001; women: R2=0.690, t=7.75, p<0.0001). No significant gender differences were found in the linear regression slope (t (150)=0.126, p>0.05).
Age-related differences in tobacco-warning images
The participants were divided in ranges of 2 years old. Because the number of participants with >23 years was reduced, these participants were all included in the same group (>23). Supplementary tables show the mean and SD ratings for each tobacco-warning image as a function of age (13–14, 15–16, 17–18, 19–20, 21–22 and >23 years old). Between 6 and 8, images were evaluated as pleasant (six images in the 15–16 and 19–20 age groups, seven images in the 13–14 and 21–22 age groups and eight images in the 17–18 and >23 age groups). For the unpleasant tobacco-warning images, all age groups showed a significant negative correlation between the ratings of valence and arousal (13–14: r=−0.799, p<0.0001; 15–16: r=−0.738, p<0.0001; 17–18: r=−0.473, p<0.05; 19–20: r=−0.767, p<0.0001; 21–22: r=−742, p<0.0001 and 23<: r=−0.772, p<0.0001) and a significant negative linear regression slope (13–14 years old: R2=0.639, t=6.77, p<0.0001; 15–16 years old: R2=0.545, t=5.69, p<0.0001; 17–18 years old: R2=0.224, t=2.684, p<0.0001; 19–20 years old: R2=0.588, t=−6.214, p<0.0001; 21–22 years old: R2=0.551, t=5.644, p<0.0001 and >23 years old: R2=0.596, t=6.064, p<0.0001) (see figure 3). Significant age differences in the linear regression slope were found between the 15–16 age group (−1.163) and the 13–14, 17–18, 19–20, 21–22 and >23 age groups (−0.837, −0.530, −0.817, −720 and −0.728, respectively, all p values <0.05).
Smoking status differences in tobacco-warning images
Participants were classified into four smoking status groups based on their responses to the smoking habits questionnaire: non-smokers (“I have never tried smoking”), one-time smokers (“I don't smoke, but I have tried smoking”), occasional smokers (“I smoke sometimes, two or three times a month”) and heavy smokers (“I usually smoke every day”). The ratings of the tobacco-warning images for each smoking status group ranged from 1.82 to 7.36 on the valence scale and from 3.18 to 6.75 on the arousal scale (see table S2 in supplementary material). Figure 4 represents the distribution of the 35 tobacco-warning images in the valence–arousal space for each smoking status group. For the unpleasant tobacco-warning images, all groups showed significant negative correlations between the ratings of valence and arousal (non-smokers: r=−0.773, p<0.0001; one-time smokers: r=−0.789, p<0.0001; occasional smokers: r=−0.736, p<0.0001 and heavy smokers: r=−0.622, p<0.0001) and significant negative linear regression slopes (non-smokers: R2=0.598, t=6.33, p<0.0001; one-time smokers: R2=0.623, t=6.55, p<0.0001; occasional smokers: R2=0.540, t=4.42, p<0.0001 and heavy smokers: R2=0.387, t=4.13, p<0.0001). No significant group differences were found in the linear regression slope.
Gender, age and smoking influences in the rating of tobacco-warning images
Results of the stepwise regression for valence ratings of unpleasant tobacco-warning images yielded a significant model (R2=0.032, F(2,591)=10.816, p<0.0001) with two predictor variables: gender (r=−0.152, p<0.001) and heavy smokers with ages between 19 and 20 (r=0.118, p<0.005). Female participants evaluated negative tobacco-warning images as more negative, whereas heavy smokers with ages between 19 and 20 evaluated the same images as less negative than the other participants. Results of the stepwise regression for arousal ratings also yielded a significant model (R2=0.180, F(5,591)=26.951, p<0.0001) with five predictor variables: gender (r=0.179, p<0.0001), one-time smokers with a range of age between 15 and 16 years old (r=0.135, p<0.002), occasional female smokers (r=−0.134, p<0.003), age range between 13 and 14 (r=−166, p<0.0001), and age range between 15 and 16 years old (r=−0.333, p<0.0001). Thus, female participants and 15- and 16-year-old adolescents who were one-time smokers evaluated negative tobacco-warning images as more arousing than the other participants. In contrast, 13- and 16-year-old adolescents and occasional female smokers evaluated the images as less arousing than the other groups.
The present evaluation of the emotional impact of tobacco-warning images applied a well-established procedure in emotion research to examine appetitive and defensive motivational tendencies evoked by images.23 The results indicate that the majority of the tobacco-warning pictures proposed by the European Commission (83%) were rated within the unpleasant cluster with scores ranging from moderately unpleasant to very unpleasant. However, a small but significant number of pictures (17%) were evaluated as positive. The positive scores ranged from moderately pleasant to very pleasant. In general, the tobacco-warning pictures were rated as moderately arousing. Only four pictures (11.4%) had scores >6 on the 1–9 arousal scale.
The distribution of these pictures alongside the affective IAPS pictures is coherent with the two hypothetical underlying motivational systems proposed by Lang and colleagues16: the appetitive (top half of the plot) and the defensive (bottom half of the plot). The boomerang-shaped distribution of the pictures shows that the pictures that were rated as either more pleasant or more unpleasant were also rated as more arousing. This finding is confirmed by the significant negative correlation (and negative linear regression) between valence and arousal found for the tobacco-warning pictures located within the unpleasant cluster. Accordingly, our results suggest that the tobacco-warning pictures proposed by the European Commission are not all rated as consistent with activation of the defensive–avoidance motivational system. Surprisingly, some pictures were rated as positive and consistent with activation of the opposite appetitive–approach motivational system. More importantly, only a very small number of unpleasant tobacco-warning pictures (11.4%) were assessed as highly arousing. If the aim of the European anti-tobacco campaign, based on aversive warning images, is to prompt negative attitudes towards smoking and predispose smokers to quit smoking by activating the defensive–avoidance motivational system, our results question the effectiveness of most of the proposed images.
Differential evaluations of the emotional impact of the tobacco-warning images by gender, age and smoking status did not substantially change the above results. All groups rated some tobacco-warning images as pleasant by locating them within the pleasant cluster, all groups showed significant negative correlations (and negative linear regression) between valence and arousal ratings for images within the unpleasant cluster and all groups rated the tobacco-warning images as moderately arousing. However, multiple regression analysis revealed significant differences in the predictive value of gender, age and smoking status on the valence and arousal ratings of the tobacco-warning images that provide insight into the potential effectiveness of the images for specific target populations.
Gender was a predictor variable for both valence and arousal ratings. Women evaluated the negative tobacco-warning images as more unpleasant and more arousing than men. Similar gender differences in ratings of affective pictures have been previously reported.29 ,31 This finding suggests that women might benefit more than men from anti-tobacco campaigns based on negative warning images because these images induce more negative and intense feelings in women. This consideration is particularly relevant if we consider that, in recent years, tobacco firms have developed specific commercial strategies addressed to persuade women to consume tobacco products, such as elegant and refined packs with slim cigarettes or packs in the form of lipstick with different colours depending on the social context.32 Therefore, unpleasant pictures on tobacco packs might be more effective for women than for men in promoting negative attitudes towards tobacco use.
Age was also a predictor variable for arousal ratings. Adolescents between 13 and 16 years old rated the tobacco-warning images as less arousing than the older groups. This finding is consistent with previous reports of increases in emotional arousal with increasing age.33 Grühn and Scheibe34 found that older adults rate negative pictures as more arousing and positive pictures as less arousing than younger adults. Thus, if younger people have a tendency to respond with less emotional intensity to negative affective pictures, tobacco-warning images may be less effective in promoting negative attitudes towards tobacco use in adolescents than in adults. Considering that young people are currently one of the main marketing targets of tobacco firms in the world,35–37 the proposed tobacco-warning images might be particularly ineffective for this target population unless more arousing unpleasant pictures are used.
Smoking status alone was not a predictor variable for valence or arousal ratings of the negative tobacco-warning images. However, in combination with age, smoking status predicted both valence and arousal ratings. One-time smokers with a range of age between 15 and 16 years old predicted higher arousal ratings, whereas heavy smokers with ages between 19 and 20 predicted less negative valence ratings. This finding is relevant because one-time smokers have more exposure to the potential influence of warning images on tobacco packs. Because an addiction is not yet established in these smokers, the more arousing unpleasant images that they see when they buy their tobacco packs might induce negative attitudes activating the defensive motivational system. In contrast, tobacco-warning images might be less effective in persuading heavy smokers to quit smoking with a range of age between 19 and 20 years old. Several studies have shown that cigarette brands are conditioned cues that evoke craving, drug seeking and tobacco use.38–40 If the tobacco-warning images are not sufficiently negative and arousing, they may fail to activate negative attitudes. Thus, more arousing and aversive images might be necessary to counteract the conditioned effects of tobacco brands in this group of smokers.37 ,41
The implications of our findings should be assessed with the consideration of some methodological limitations. First, our evaluation was limited to the tobacco-warning images per se, that is, without the context of the actual tobacco package. Second, no warning texts or symbols were included with the images, whereas the European-recommended warnings include both images and text. Third, the participants were unevenly distributed among the different subgroups, which prevented an analysis of potential interactions between gender, age and smoking status. Moreover, all participants were asked about their smoking behaviour after seeing the stimuli. It is possible that participants guessed that the study was about tobacco use and that this may have influenced their responses to the tobacco use questions. All of these variables are relevant factors that future research should examine to extend the scope of the present study.
Subject to the limitations above, the present results suggest that the warning images proposed by the European Commission for tobacco packages might have limited effectiveness in reducing tobacco consumption in the general population because most of the proposed images were evaluated as moderately unpleasant and arousing. Because such images may not be capable of inducing negative attitudes and avoidance behaviours, the question of their effectiveness remains open. However, the differences found in the intensity of the impact depending on gender, age and smoking status allow for the conclusion that the proposed warning images might be effective in specific target populations, such as women, adults (over 17 years old) and one-time young smokers. New images, well selected for their high negative valence and arousal, might be necessary to increase the effectiveness of anti-tobacco campaigns based on unpleasant warning images.
What this paper adds
First evaluation of the emotional impact of European tobacco-warning images in a wide sample of participants including adolescents and adults.
Comparison of the European tobacco-warning images with standard pleasant, neutral and unpleasant images from the International Affective Picture System.
The results question the potential effectiveness of the images in the general population, unless more arousing unpleasant images are used.
The finding of differential impact of the images depending on gender, age and smoking status suggests potential effectiveness in specific target populations.
We thank HUM-388 research group (University of Granada), I.E.S José Rodrigo-Botet (Valencia) and Col·legi Concertates Liceu (Islas Baleares) for their support to accomplish this research.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
Funding This work was supported by Spanish Ministry of Science and Innovation, grant number JCI-2008-3074.
Competing interests None.
Patient consent Obtained.
Ethics approval Ethical committee of University of Granada and ethical committee of University of Balearic Islands.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Data might be available only on request to MAM.