A power calculation performed using the G*power software suggests that at least 52 subjects per condition are needed to detect a medium effect size (most of effect sizes for WTP were larger than 0.3 in Simmonds et al., 2018b) of d = 0.8 with α = 0.05 for a one-way ANOVA analysis. In addition, a recent review on eye-tracking studies of nutrition label processing indicated that a small and convenient sampling could provide reliable findings on consumer behavior (Ma and Zhuang, 2020). In the review, there were 33% studies with sample size less than 60 participants and the smallest sample size was only 10 participants (Bialkova and van Trijp, 2011). Therefore, a convenient sample of 55 college students (30 females, ages between 18 and 21; 25 males, aged between 18 and 21) from Shaanxi Normal University in Xi ‘an, China, was paid to participate in the experiment. The participants had no color blindness, color weakness, or history of mental illness. They had normal or corrected-to-normal vision. Ten students were unable to conduct the formal experiment due to calibration or other eye-tracking problems. Thus, 45 participants (25 females and 20 males) were kept for final data analyses.

A pilot study was performed before creating the experimental materials. The pilot study investigated the popularity and category of products with transparent packaging in markets in Xi’an. Three off-line supermarkets (Vanguard, Wal-Mart, Yonghui) were visited to survey the transparent materials used to package food products. The survey found that the packaging materials of prepackaged food on the shelves were divided into three categories: box, bag, and can. Of the 170 food brands surveyed, 70.6% of them used bag packaging, 17.1% packaged their products in cans, and 12.3% of them used boxes. Since bags are the most commonly used packaging in prepackaged foods, all food packaging pictures in this study were designed on bags. The pilot study also showed that transparent packaging was frequently used for biscuits, cereals, grain products, potatoes and puffed grain foods, candy products, preserved fruits and other fruit products, stir-fried food, nut products, and frozen foods (see Figure 1). Moreover, our online survey based on a large e-commerce platform in China showed that nut products, preserved fruits, and instant cereals were the three most popular foods for sale. Thus, these three categories were selected to create the experimental materials.

Each product category contains nine different foods, forming a total of 27 food products included in the current study. Adobe Photoshop CS6 software was used to create pictures of experimental food packaging. Each packaging picture contains elements such as the trademark, food name, a picture or transparent window, certification mark, and food net content. Each food was packaged in three different ways: transparent window packaging (transparent), graphic window packaging (graphic), and baseline window packaging (baseline). To avoid the influence of food trademarks on the subjects, all foods were named with a nonexistent brand that the subjects had never seen before (Simmonds et al., 2018b). In order to reduce the visual differences of the three versions of packaging, we matched the position and size of food trademarks and windows, which were also based on actual products in supermarkets to increase ecological validity. Finally, 81 food images were created as stimuli in the formal experiment (see Figure 2 for an illustration of experimental stimuli).

The stimuli were presented on a 24-inch LCD monitor (ASUS VG248QE) with a resolution of 1920 × 1080 pixels and a refresh rate of 144 Hz. The distance between the participants’ eyes and the screen was 62 cm. Each packaging image occupied about 24 cm vertically and 20 cm horizontally on the screen. Participants’ eye movements were recorded by an EyeLink 1000 Plus (SR Research Ltd., Ontario, Canada) eye tracker with a sample rate of 1,000 Hz. A chin rest was used to reduce head movements. Since both eyes fixate on the same spot, recording one eye is sufficient. Experimental data were collected and processed by the Experiment Builder and Data Viewer software.

After arriving at the lab, participants were asked to rate their hunger levels (1 = not hungry at all, 7 = very hungry) and their desire to eat (1 = not strong at all, 7 = very strong) using a seven-point Likert scale. If a participant’s score was higher than 4, they were asked to eat snacks that were unrelated to the experimental stimuli. When asked again, the participant rated their hunger and desire to eat as less than 4 after eating the snacks. Single factor Latin square experimental design was adopted for the three package types (i.e., transparent window packaging, graphic window packaging, and baseline window packaging). There were three sets of trials, and each set contained 27 trials, which included different food packaging. Participants were randomly assigned to each set.

During the experimental stage, the researcher briefly introduced the overall procedure and experimental equipment to the participants, and then carried out a five-point calibration. After successful calibration, participants finished three practice trials to familiarize the experimental procedure. None of the stimuli used in the practice sessions appeared in the formal experiment. The participants then performed the formal experiment with 27 trials (see Figure 3 for a demonstration). After participants focused on a fixation cross for 300 ms, the packaging images appeared, and participants were asked to view the images freely, pressing the space bar when they were ready to proceed to the next stage. Participants were then asked to estimate their willingness to purchase the product (i.e., How likely would you be to buy this food, assuming it was available and at a reasonable price?) and the product’s visual attractiveness (i.e., How attractive is the packaging to you?) of each packaging image, using seven-point Likert scales.

The main region of interest (ROI) is the window region for each type of packaging. Dependent variables included two perceptual measures (willingness to purchase and packaging attractiveness), and the following three eye movement measures for each ROI. Time to first fixation (TTFF; i.e., latent time to first fixate an interest area Ma and Zhuang, 2020) was selected as one of the main indicators of attention capture by salient features. The shorter the time to the first fixation was, the more the salience effect was apparent for the observed area of interest. We also selected two most commonly used eye movement measures: total time (TT; i.e., the sum of the duration across all fixations that fall in the current interest area) and number of fixations (NF; i.e., total number of fixations falling in the interest area) to reveal the total attention interest in the corresponding region (Wedel and Pieters, 2007; Holmqvist et al., 2011). The extreme value (±2.5 SD) of the concerned eye movement measures was excluded. More than 97.6% of data were kept for final analyses. ANOVA was used to analyze the variance of all dependent variables, and Bonferroni corrections were used for paired comparisons. Spearman’s correlation coefficient was used to test the correlation between the eye movement indexes and behavior indexes with SPSS 22.0.

In previous studies, according to the salience effect hypothesis, researchers proposed that packaging with transparent elements would always stand out to consumers compared with opaque packaging, and attracted consumers to eat more, or promoted their purchase behavior (Deng and Srinivasan, 2013; Simmonds et al., 2018b). However, using the eye-tracking approach is necessary to definitively ascertain if packaging with transparent elements really captures consumer attention. Generally, consumers tend to look at the top half of a novel product’s packaging (Juravle et al., 2015). However, even a potentially preset scanpath does not influence the comparison of packaging attraction for different packaging types in the current within-subject design experiment.

Time to first fixation was used as one of the main indicators to investigate the salience effect in this study. The data showed that only preserved fruits had a significant main effect and whose transparent window packaging successfully attracted consumer attention. For nuts and instant cereals, however, even the descriptive statistics did not show a shorter time to first fixation for product packaging with a transparent window. These data were partially inconsistent with the prediction of the salience effect hypothesis (Spence et al., 2016), which might be caused by the food categories we selected. The foods in this study were selected according to the utilization rate of transparent packaging and food sales volume, which meant that participants had corresponding purchasing experience or a personal preference for these foods. By involving personal expectations, their gaze behaviors were not only affected by the stimulus’ physical characteristics (bottom-up) but also by the participants’ subjective expectations (top-down). Therefore, the early-stage measure of time to first fixation may reflect the interaction between the participants’ bottom-up and top-down processing of the packaging pictures.

The later-stage measures, such as total time and number of fixations, always reflected sustained attention. A number of studies have shown that visually salient stimulus not only captured participants’ attention quickly but also retained longer processing time (Shimojo et al., 2003; Bialkova and van Trijp, 2011; Milosavljevic et al., 2012; Orquin and Mueller Loose, 2013). In this study, total time and number of fixations of the transparent and graphic window packaging of all three food categories were higher than the baseline window packaging. Also, there was no statistically significant difference between the transparent and graphic window packaging in any food category, except the number of fixations in nuts packaging. These results indicate that showing food information, whether the actual food, or a picture of it, can stand out and retain consumers’ attention, thus H1 is partially verified. Previous studies emphasized the salience effect of transparent packaging but ignored the effects of graphic packaging, which may be related to experimental materials. Deng and Srinivasan (2013) did not involve the category of graphic window packaging in their study. Simmonds et al. (2018b) included graphic window packaging, but they only selected one product out of each food category, thus the generalizability of the transparent packaging effect needs to be further examined. The materials used in the present study consist of nine types of food for each food category, which can provide higher ecological and statistical validity. The data suggest that salience is not limited to transparent window packaging because food-related graphic window packaging can also capture consumers’ attention.

Shimojo et al. (2003) proposed a dual-contribution model in which cognitive assessment systems and orienting behavior structures simultaneously influence people’s preferential decision-making behavior, through facial attractiveness decision experiments. When a stimulus attracts our attention and produces a preference to gaze it, the preference will lead to more exposure to the stimulus, which creates an increased preference. Preference in turn increases our gaze behavior, so as to continuously strengthen our perception of stimulus to influence our decisions (Shimojo et al., 2003; Simion and Shimojo, 2006). This model provides a plausible explanation for the mechanism underlying the transparent window salience effect. The explanation is that when consumers are exposed to the food through a transparent window, the food captures our attention, which seemly produces a gaze preference for the transparent window packaging compared with other packaging. This preference may make consumers pay more attention to transparent window packaging, and the increase in gaze time translates into a higher preference, which could lead to increased willingness to purchase the food in that packaging. Therefore, total time and number of fixations should have a significant positive correlation with the willingness to purchase. However, the results of this study show that only total time and number of fixations of preserved fruits were significantly and positively correlated with purchase intention. For nuts and instant cereals, none of the correlations between fixation patterns and willingness to purchase were statistically significant. These results were inconsistent with H2. This means that long gaze duration does not necessarily lead to a final purchase decision (see also Balcombe et al., 2017; Wolf et al., 2019). Wolf et al. (2019) found that the relationship between gaze time and three options [i.e., the exclusive evaluation task contains rejection, deferment, and inclusion; the non-exclusive evaluation task contains 1 (“not at all”), 2 (neutral), and 3 (“very much”)] from participants who presented an inverted U-shaped trend. Therefore, the gaze may not necessarily lead to more liking but contribute to the evaluative processing by integrating extra information. As Wolf et al. (2019) suggested, the prolonged viewing time for the middle category may reflect doubt or uncertainty during the evaluative processing, potentially with an increased effort of information integration before reaching a conclusion.

In this study, there was a significant positive correlation between perceived packaging attractiveness and a willingness to purchase. After the formal experiment, when participants were asked why they did or did not want to buy the product, almost all of them mentioned their impressions of the packaging. We speculate that consumers’ willingness to purchase is likely to be affected by the aesthetic perception of packaging. The correlation analyses also revealed that there was a significant positive correlation between packaging attractiveness and willingness to purchase. Simmonds et al. (2018b) proposed that the packaging attractiveness was a necessary premise of supporting the salience effect of transparent window packaging. The authors argued that transparent window packaging could enhance consumers’ purchase intention by highlighting the food inside, and attractive transparent window packaging can make this effect more notable. However, the results of the current study suggest that packaging attractiveness appears to work at a more general level. Showing the food in a concrete form (i.e., presenting actual foods through transparent windows or displaying food images on graphic windows) to consumers can indeed attract consumers’ attention effectively. However, the attractiveness of food packaging most significantly promotes consumers’ willingness to purchase the product.

Combined with eye movement and perceptual measures, this study further revealed that product category regulated the effect of packaging type on consumer attention. That means consumers may have different gaze patterns for different products. For nuts, most are shelled, and the difference between the actual food inside the packaging and the food image is slight. The data showed that there were no statistical differences between “transparent” and “graphic” stimuli for all the measures except for number of fixations, which suggests that both transparent and graphic window packaging have similar salience effects for nuts. Therefore, transparent window packaging and graphic window packaging can efficiently capture consumers’ attention and increase product attractiveness.

Participants preferred the graphic window packaging on preserved fruits more than the transparent window packaging according to the perceptual results, although the preference was slight. In combination with participants’ feedback after the experiment, we found that most types of preserved fruits in this study were often coated in sugar or honey. These sticky coatings would smear against the inner transparent window of the packaging and reduce participants’ appetite. Therefore, when participants saw the sugar‐ or honey-coated preserved fruits through the transparent window, they would avoid the package. Most preserved fruits are sticky or frosted, and thus graphic window packaging is more suitable to attract consumers to these products.

In their study, Simmonds et al. (2018b) also inspected instant cereals. Although the pictures of the cereal packaging in this study were similar to their experimental materials, there are several differences. Only one food for each food category was used in their study, and the baseline was a blank condition. In this study, nine types of foods for each category were selected, and the baseline was size-matched non-food graphic objects. Moreover, Simmonds et al. (2018b) asked participants to sort different packaging types presented at the same time according to participants’ judgments. This method was more likely to induce significant differences between the transparent and graphic conditions than the one-by-one estimation method used in the present study. These differences may contribute to the different conclusion they arrived at, where the salience effect only applied to transparent conditions. The present study revealed salience effects for both transparent and graphic conditions.

The present study has several limitations. First, all conclusions of this study are based on Chinese college students and packaging types used in the Chinese market. Future research could explore the salience effect in food packaging across varied age groups and different cultures. Second, this study only involves three food categories, and all of them belong to “leisure food.” Future studies can enrich and subdivide food categories (such as fresh food and cooked food) to investigate the generalization of the salience effect. Third, whether the crown in baseline packaging played as a novel stimulus in a way for some participants was not clear, which should be a more concern in future researches. Fourth, this study adopts traditional laboratory methods, which may be not enough to measure ecological validity. Besides, participants in the current study had known their task to assess willingness to purchase, which might drive their eye movement behavior. Future studies should use more ecological tasks as that in real-life shopping. Finally, in reality, transparent packaging not only allows the real food to be seen through a clear window but also allows consumers to get a full view of the food by shaking the package. However, the latter function is lost in the transparent packing images when viewed on a screen. This may be one of the reasons that diminished the advantages of transparent packaging in this study. Thus, researchers can use field experiments or VR technology to further improve the ecological validity of corresponding studies.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.