Abstract
Unhealthy diets are widespread and linked to a number of detrimental clinical outcomes. The current preregistered experiment extended expectancy theory into the study of food intake; specifically, we tested whether a fast-food restaurant affects food expectancies, or the emotions one expects to feel while eating highly processed foods (e.g., pizza) and minimally processed foods (e.g., carrots). Participants (N = 200, mean age = 18.79 years) entered a simulated fast-food restaurant or a neutral space, completed questionnaires, and engaged in a bogus taste test. The simulated fast-food restaurant increased positive highly processed food expectancies (d = 0.29). Palatable eating coping motives scores did not moderate the effect; however, this clinically relevant pattern of eating behavior was associated with greater positive highly processed food expectancies. In addition, there was an indirect effect of the fast-food restaurant on ad libitum food intake through positive highly processed food expectancies. Reducing positive highly processed food expectancies may improve diet, which may broadly affect health.
Keywords
Unhealthy diets are widespread. Average people in the United States get the majority of their energy intake from highly processed foods or processed foods that are high in refined carbohydrates and fats (Martinez Steele et al., 2016). A diet that largely consists of highly processed foods at the expense of minimally processed foods (e.g., fruits, vegetables) is associated with a number of clinical outcomes, including all-cause mortality, overall cardiovascular diseases, coronary heart diseases, cerebrovascular diseases, hypertension, metabolic syndrome, depression, irritable bowel syndrome, overall cancer, and postmenopausal breast cancer (Chen et al., 2020). Moreover, an estimated 15% of the U.S. population experiences an addictive-like response to highly processed foods, including subjective experiences of impaired control, cravings, tolerance, and withdrawal (Schulte & Gearhardt, 2018). It is therefore imperative to understand the mechanisms guiding food intake, particularly of highly processed options. Extending theory from psychological science into the study of food intake may provide novel insight.
Expectancy theory is a long-standing psychological theory explaining behavior (W. James, 1890/2007; Tolman, 1949). It proposes that people learn from personal or vicarious experiences what outcomes a behavior can lead to, including positive and negative emotions. These expectancies influence the likelihood of engaging in the behavior (Bandura, 1977). For instance, a person may learn from personal experiences or other people that alcohol can make people feel happier; positive alcohol expectancies encourage a person to drink more frequently and heavily (Goldman et al., 1999). Psychologists have predominantly applied expectancy theory to explain substance use behaviors, including alcohol, cigarette, and marijuana use (Buckner & Schmidt, 2008; Cohen et al., 2002; Morean et al., 2012), and disordered eating, including symptoms of anorexia and bulimia (Culbert et al., 2015; Pearson et al., 2014). More recently, Cummings, Joyner, and Gearhardt (2020) developed a measure of food expectancies, or the positive and negative emotions one anticipates feeling while eating highly processed (e.g., pizza) and minimally processed (e.g., carrots) foods. Food expectancies—in particular, positive highly processed food expectancies (e.g., anticipating one will feel happy while eating pizza)—were associated with less healthy patterns of food intake among adults.
Expectancy theory posits that expectancies form early in development, dynamically change during adolescence, and become relatively stable in young adulthood (Smith & Goldman, 1994). However, the situational-specificity hypothesis from expectancy theory suggests that environments can temporarily distort expectancies about environment-relevant behavior (Wall et al., 2000). For example, when college students were in a bar as opposed to a neutral space, they were more likely to anticipate feeling “happy” and “outgoing” from drinking alcohol (Wall et al., 2000). In addition to increasing positive alcohol expectancies, one study found that being in a bar had a nonuniform effect on negative alcohol expectancies: When college students were in a bar as opposed to a neutral space, they were less likely to anticipate feeling “clumsy” and “slow” but were more likely to anticipate they would “act aggressively” and “be loud, boisterous, or noisy” from drinking alcohol (Wall et al., 2001). However, two other studies found that being in a bar increased only positive alcohol expectancies (Monk & Heim, 2013; Wall et al., 2000). Overall, empirical support for the situational-specificity hypothesis supports the notion that environments acutely increase, at least, positive expectancies about environment-relevant behavior.
No study (to our knowledge) has tested whether an environment can temporarily distort food expectancies. One major environment that may affect food expectancies is a fast-food restaurant, or a limited-service restaurant that quickly prepares food for individuals to eat at the restaurant or to bring elsewhere (P. James et al., 2014). An estimated 78% of U.S. ZIP codes have at least one fast-food restaurant (Powell et al., 2007). In particular, fast-food restaurants almost exclusively offer highly processed foods with a very limited number of minimally processed foods (Kirkpatrick et al., 2014). Menus from fast-food restaurants often describe highly processed foods with positive emotion words (e.g., McDonald’s “happy” meal; Wendy’s “S’Awesome” bacon cheeseburger). Advertisements for fast-food restaurants make up about one third of food advertisements on general market television in the United States (Henderson & Kelly, 2005), and marketers construct food advertisements so viewers associate emotions with the food advertised (Page & Brewster, 2009). In addition, fast-food restaurants include pleasant stimuli like music, brightly colored menus, and brightly colored furniture as well as food cues like food smells and sight of other patrons’ food, which patrons may associate with the highly processed foods offered there—albeit other types of restaurants include these pleasant stimuli and food cues too. A fast-food restaurant is therefore a candidate environment for acutely changing food expectancies (at least increasing positive highly processed food expectancies) because of the types of food it offers and because it houses food descriptions, an array of stimuli, and food cues.
It is also possible that certain individuals may be more susceptible to environmental effects on expectancies, yet this has not been tested in prior work on the situational-specificity hypothesis (Monk & Heim, 2013; Wall et al., 2000, 2001). In the context of fast-food restaurants, which create positive emotional experiences around eating highly processed foods, food expectancies may especially be affected among individuals who typically eat highly processed foods for emotional reasons. This notion is supported by prior work showing that individuals who eat for emotional reasons are more reactive to laboratory mood inductions (van Strien et al., 2012, 2013)—albeit different laboratory groups have found inconsistent effects (for a review, see Bongers & Jansen, 2016). Focusing on individual differences in emotional eating when investigating food expectancies holds clinical relevance because emotional eating is associated with clinical outcomes such as metabolic syndrome, diabetes, and depression (Epel et al., 2004; Ouwens et al., 2009; Tsenkova et al., 2013). Among college students, in particular, a one-point change in scores on the Coping Motives subscale of the Palatable Eating Motives Scale (PEMS; i.e., a specific pattern of eating highly processed foods to cope with negativity) predicted an average weight gain of 10.5 lbs over 2 years (Boggiano, Wenger, Turan, Tatum, Morgan, & Sylvester, 2015).
Furthermore, if expectancies have a causal influence on behavior, then temporary distortions in expectancies caused by the environment should subsequently affect environment-relevant behavior (Bandura, 1977). However, prior work investigating the situational-specificity hypothesis has not tested this critical tenet of expectancy theory (Monk & Heim, 2013; Wall et al., 2000, 2001). Identifying whether temporarily distorted food expectancies affects food intake is not only theoretically important but also practically important. Successful substance use and disordered eating interventions have targeted substance and thinness expectancies, respectively, which suggests that food expectancies could be an important target for dietary interventions (Annus et al., 2008; Scott-Sheldon et al., 2012). Without evidence that food expectancies causally affect food intake, however, there lacks support for the development of such interventions.
To fill the gap in the existing literature, the primary preregistered aim of the current study was to test the effect of a simulated fast-food restaurant on food expectancies. This environment was used in a prior study on the effects of a food-cue-rich environment on food intake (Joyner et al., 2017). In accordance with the situational-specificity hypothesis, the preregistered hypothesis was that exposure to this simulated fast-food restaurant would increase positive highly processed food expectancies. Given mixed findings regarding the effect of a bar on negative alcohol expectancies (Monk & Heim, 2013; Wall et al., 2000, 2001), there was no preregistered hypothesis about the effect of the simulated fast-food restaurant on negative highly processed food expectancies (e.g., anticipating one will feel down while eating pizza). In addition, because fast-food restaurants predominantly offer highly processed foods with fewer minimally processed foods (Kirkpatrick et al., 2014), there were no preregistered hypotheses about the effect of the simulated fast-food restaurant on positive minimally processed food expectancies (e.g., anticipating one will feel happy while eating carrots) or negative minimally processed food expectancies (e.g., anticipating one will feel down while eating carrots).
To understand whether there may be individual differences in environmental effects on food expectancies, a secondary preregistered aim of the current study was to examine whether palatable eating coping motives scores moderated the effect of the simulated fast-food restaurant on food expectancies (Burgess et al., 2014). Finally, to test the causal influence of food expectancies on food intake, another secondary preregistered aim was to test for an indirect effect of the simulated fast-food restaurant on ad libitum food intake through food expectancies. Ad libitum food intake was observed through a bogus taste test—a valid method for measuring actual food intake without the limitations of self-report methods (Robinson et al., 2017).
Method
Design
The study design was a two-level (fast-food restaurant vs. neutral space) randomized between-subjects experiment.
Participants
Prior research observed a medium effect (η2 = .06) of the simulated fast-food restaurant on food “wanting” (Joyner et al., 2017). A priori power analysis in G*Power (Version 3.1.7; Faul et al., 2007) indicated a sample size of 144 for the current study on the basis of the following parameters: one-way analysis of variance (ANOVA), η2 = .06, α = .05, .85 power, and two groups. We aimed to recruit at least 144 participants from the University of Michigan’s Department of Psychology subject pool; however, to further increase power, to mitigate concern of participants who would be excluded, and because resources were available to continue supporting the study, we continued recruiting participants through the end of the academic semester for fall 2019. In total, we recruited 203 participants. Following the preregistration plan, we excluded three participants because of researcher error (i.e., research assistant did not administer the Anticipated Effects of Food Scale [AEFS], n = 1) or participant ineligibility (i.e., participant requested data be destroyed, n = 1; participant verbally admitted to intoxication during study, n = 1). The final sample thus comprised 200 participants. Table 1 presents demographic and behavioral characteristics of the sample overall and by experimental condition.
Demographic and Behavioral Characteristics of the Sample Overall and by Experimental Condition
Note: Values in parentheses are standard deviation unless otherwise noted. BMI = body mass index; PEMS = Palatable Eating Motives Scale.
Procedure
Participants were recruited to participate in a study in which they would report on their beliefs about food, alcohol, and other drugs. The university institutional review board approved the research procedure in accordance with the provisions of the World Medical Association Declaration of Helsinki. Sessions were scheduled Monday through Friday between 1 p.m. and 5:30 p.m.
After providing written informed consent, participants were randomly assigned to enter the simulated fast-food restaurant or a neutral space (control). The simulated fast-food restaurant included the following characteristics: brightly colored booths, dining tables and chairs, and visible industrial-style food storage and preparation appliances through a kitchen window (Joyner et al., 2017). Before participants arrived at the lab, research assistants cooked French fries in the kitchen to create food smells that would last for participants’ visits. For photo images of the simulated fast-food restaurant, see Figure S1 in the Supplemental Material available online. We selected an office space for the neutral space because people occasionally eat at the office, but it is not a food-cue-rich environment (Oh et al., 2014). The neutral space included the following characteristics: office chairs, office desks, filing cabinets, a printer, and a desktop computer. The walls, carpet, and furniture had neutral colors.
Participants seated themselves in the environment they were randomly assigned to and then completed an initial battery of questionnaires in randomized order, including the AEFS. Hunger, taste ratings of foods, and ad libitum food intake were next measured during a bogus taste test paradigm (Robinson et al., 2017). Participants were instructed to taste and rate one Oreo mini cookie (from ~25), one Lay’s plain potato chip (from ~20), one baby carrot (from ~20), and one grape (from ~25) in their preferred order. Then, participants were left alone for 5 min to help themselves to the remaining cookies, potato chips, baby carrots, and grapes while research assistants set up for the next task. Participants were also given a bottle of water (8 fluid oz).
Following the bogus taste test paradigm, participants completed a second battery of questionnaires in randomized order, including PEMS and “liking” of food items. Participants then reported on when they last ate before the study and their demographics. Research assistants measured participants’ heights with a stadiometer and weights using an InBody 570 (Cerritos, CA). Finally, participants were debriefed and compensated with course credit. Research assistants rated the participants’ levels of suspicion of the study’s true purpose as an average 1.71 (SD = 0.79) out of 5 (1 = no suspicion, 5 = very suspicious).
Measures 1
Anticipated Effects of Food Scale
The AEFS (Cummings, Joyner, & Gearhardt, 2020) is a 62-item questionnaire that measures the anticipated positive and negative emotional outcomes of eating highly processed foods (labeled as “junk food” to improve readability) and minimally processed foods (labeled as “healthy food” to improve readability). In the assessment of highly processed food expectancies, the instructions were as follows: “Imagine that you are eating JUNK food (e.g., sweets, salty snacks, fast foods, sugary drinks) . . . how much do you expect to feel the following feelings
In the assessment of minimally processed food expectancies, the instructions were as follows: “Imagine that you are eating HEALTHY food (e.g., fruits, vegetables) . . . how much do you expect to feel the following feelings
Hunger, taste ratings of foods, and ad libitum food intake
At the beginning of the bogus taste test paradigm, participants rated their level of hunger on a visual analogue scale from 0 (not hungry at all) to 100 (it’s all I can think about). After trying one of each food item, participants rated how much they liked the taste of the food on a visual analogue scale from −100 (extremely dislike) to 100 (extremely like). Before and after the bogus taste test paradigm, research assistants weighed the Oreo mini cookies, Lay’s plain potato chips, baby carrots, and grapes in grams. We chose restaurant-inconsistent foods because prior work demonstrated the simulated fast-food restaurant increased intake of restaurant-consistent foods (i.e., cheeseburgers, fries; Joyner et al., 2017). Choosing restaurant-inconsistent foods therefore allowed us to increase detection of an indirect effect (whether food intake differed between groups because of food expectancies) in addition to the direct effect of the simulated fast-food restaurant on food intake. Choosing restaurant-inconsistent foods also allowed us to test whether the direct effect of the simulated fast-food restaurant would generalize to restaurant-inconsistent foods.
Nutritional information for each food item is provided in Table S1 in the Supplemental Material. To index ad libitum food intake, we first subtracted the postexperiment weight of each food item from the preexperiment weight, and we multiplied the results by the number of calories per gram in the food item (provided on the labeled nutrition facts). We next categorized the number of calories eaten from the available highly processed foods (i.e., Oreo mini cookies and Lay’s plain potato chips) and minimally processed foods (i.e., baby carrots and grapes). Finally, we divided the number of highly processed food calories eaten by the number of minimally processed food calories eaten to create a ratio. A higher ratio indicated that a participant ate a greater number of calories from highly processed foods at the expense of calories from minimally processed foods. We chose to index ad libitum food intake this way on the basis of trends in epidemiological research on diet and health (Kennedy et al., 1995; Monteiro et al., 2018). These trends recognize that indexes that simultaneously account for multiple types of food intake may be more clinically meaningful than indexes that capture only one type of food intake.
Palatable Eating Motives Scale
The PEMS (Burgess et al., 2014) measures different reasons for eating highly processed foods; for the current study, we focused on the PEMS Coping Motives subscale (i.e., a pattern of eating highly processed foods to cope with negativity). Participants were given examples of highly processed foods adopted from the Yale Food Addiction Scale (Gearhardt et al., 2009) and then asked to report how often they ate these foods for each reason included in the four-item subscale. Sample items from the PEMS Coping Motives subscale include “to forget your worries” and “because it helps you when you feel depressed or nervous.” Participants rated items on a 5-point Likert scale from 1 (almost never/never) to 5 (almost always/always). The PEMS Coping Motives subscale demonstrated good internal consistency (α = .88).
Liking of foods, when participants last ate, and demographics
Participants reported how much they generally like Oreo mini cookies, Lay’s plain potato chips, baby carrots, and grapes on a visual analogue scale from −100 (extremely dislike) to 100 (extremely like). Participants reported when they last ate (0–1 hr previously, 1–3 hr previously, 3–5 hr previously, 5–7 hr previously, or > 7 hr previously), their age, their biological sex assigned at birth, their race/ethnicity, and the highest level of their parents’ education. We calculated body mass index (BMI) from height and weight measurements using the standard formula (kg/m2).
Data analytic plan
Following the preregistration plan, we assessed all variables for normality. Age and body mass index (BMI) showed evidence of skew (> 1) and kurtosis (> 3), and we thus log-transformed those variables for analysis. We created dummy codes for biological sex (0 = male, 1 = female) and race/ethnicity (0 = non-Black, 1 = Black); we dummy coded race/ethnicity this way because the prevalence of obesity is substantially higher in Black populations than in White populations in the United States (Hales et al., 2018; Ogden et al., 2012). To identify covariates, we conducted bivariate correlations between food expectancies and levels of suspicion, hunger, taste ratings of foods, liking of foods, when participants last ate, and demographics. Positive highly processed food expectancies were correlated with biological sex (r = −.15, p = .037) such that male participants had greater levels (M = 3.07, SD = 0.65) compared with female participants (M = 2.88, SD = 0.59). Greater positive highly processed food expectancies were correlated with greater hunger (r = .19, p = .009).
To examine the primary aim of the current study, one-way ANOVAs were conducted. Food expectancies were the dependent variables. Experimental condition was the between-subjects factor. To examine the secondary aims of the current study, multiple regressions were conducted predicting food expectancies from palatable eating coping motives and their interaction with the experimental condition (environment was dummy coded such that 0 = neutral space, 1 = fast-food restaurant). We used the PROCESS Model 4 macro (Hayes, 2013) to test the indirect effect of the experimental condition on ad libitum food intake through food expectancies. We used 10,000 bootstrap samples to create 95% bias-corrected and accelerated (BCa) confidence intervals (CIs) to test the significance of indirect effects. Indirect effects are significant at p < .05 if the 95% CI does not include zero. We present unadjusted estimates from analysis conducted without covariates in the Results section below, and we present adjusted estimates from analysis conducted with covariates in the Supplemental Material. We conducted all analyses in IBM SPSS (Version 25).
Results
Table 1 presents estimates showing that there were no differences in demographic and behavioral characteristics by experimental condition and no differences in hunger by experimental condition.
Primary aim
Table 2 presents group means and standard deviations for food expectancies and unadjusted estimates from analysis. Participants in the simulated fast-food restaurant reported greater positive highly processed food expectancies compared with participants in the neutral space. Participants in each environment did not significantly differ on their reports of negative highly processed food expectancies, nor did they differ on positive and negative minimally processed food expectancies. Including hunger and biological sex as covariates in the model did not change the direction, magnitude, or significance of these results, and there was no significant interaction between environment and biological sex on food expectancies (for adjusted estimates, see Table S2 in the Supplemental Material).
Means and Standard Deviations of Food Expectancies Overall and by Experimental Condition
Note: Values for conditions are means with standard deviations in parentheses. +HPF = positive highly processed food; -HPF = negative highly processed food; +MPF = positive minimally processed food; -MPF = negative minimally processed food.
Secondary aims
Table 3 presents bivariate correlations among food expectancies, palatable eating coping motives, and ad libitum food intake.
Correlations Among Food Expectancies, Palatable Eating Coping Motives, and Ad Libitum Food Intake
Note: +HPF = positive highly processed food; -HPF = negative highly processed food; +MPF = positive minimally processed food; -MPF = minimally processed food; PEMS = Palatable Eating Motives Scale; HPF = highly processed food; MPF = minimally processed food.
p < .05. **p < .01. ***p < .001.
Palatable eating coping motives
Palatable eating coping motives scores did not moderate the effect of the simulated fast-food restaurant on food expectancies (ps > .33). Irrespective of experimental condition, greater palatable eating coping motives scores predicted greater positive highly processed food expectancies (unadjusted b = 0.16, SE = 0.05, p = .001, 95% CI = [0.06, 0.25]), negative highly processed food expectancies (unadjusted b = 0.22, SE = 0.06, p < .001, 95% CI = [0.10, 0.33]), and negative minimally processed food expectancies (unadjusted b = 0.15, SE = 0.03, p < .001, 95% CI = [0.08, 0.21]). Palatable eating coping motives scores did not predict positive minimally processed food expectancies (unadjusted b = 0.06, SE = 0.06, p = .33, 95% CI = [−0.06, 0.17]). Including hunger and biological sex as covariates in these models did not change the direction, magnitude, or significance of these results (for adjusted estimates, see Table S3 in the Supplemental Material).
Ad libitum food intake
Figure 1 presents the unadjusted mediation model with a, b, and c′ path estimates from PROCESS Model 4. The indirect effect of the simulated fast-food restaurant on ad libitum food intake through positive highly processed food expectancies was significant (unadjusted b = 0.14, SE = 0.08, 95% BCa CI = [0.02, 0.36]). Specifically, the simulated fast-food restaurant increased positive highly processed food expectancies compared with control, which in turn caused participants to eat a greater number of calories from highly processed foods at the expense of calories from minimally processed foods. There were no indirect effects on ad libitum food intake through negative highly processed, positive minimally processed, and negative minimally processed food expectancies. Including hunger and biological sex as covariates in the model did not change the direction, magnitude, or significance of these results (for adjusted estimates, see Fig. S2 in the Supplemental Material).
PROCESS Model-4 path estimates from testing the indirect effect of the simulated fast-food restaurant on ad libitum food intake through food expectancies. Unstandardized coefficients and standard errors are presented. A dummy code was created for environment (0 = neutral space, 1 = fast-food restaurant). +HPF = positive highly processed food, -HPF = negative highly processed food, +MPF = positive minimally processed food, -MPF = negative minimally processed food. Asterisks represent significant paths (*p < .05, **p < .01).
Discussion
This is the first study (to our knowledge) to test the effect of an environment on food expectancies, to test moderation effects, and to test whether acute changes in expectancies as a result of the environment cause changes in environment-related behavior. In accordance with the preregistered hypothesis, the simulated fast-food restaurant increased positive highly processed food expectancies (e.g., anticipating one will feel happy while eating pizza). These results are consistent with prior studies showing a bar increased positive alcohol expectancies (Monk & Heim, 2013; Wall et al., 2000, 2001). They also build evidence for the situational-specificity hypothesis from expectancy theory, which suggests environments temporarily distort expectancies about environment-relevant behavior (Wall et al., 2000).
Prior work found inconsistent evidence for the effect of a bar on negative alcohol expectancies (Monk & Heim, 2013; Wall et al., 2000, 2001). In the current study, the simulated-fast food restaurant did not affect negative highly processed food expectancies (e.g., anticipating one will feel down while eating pizza). It is thus possible that environments can distort only positive expectancies about environment-relevant behavior. However, fast-food restaurants (like bars) foster positive emotional experiences. Environments fostering negative emotional experiences around eating highly processed foods (e.g., a physician’s office) may have different effects. Likewise, the simulated fast-food restaurant did not affect positive or negative minimally processed food expectancies (e.g., anticipating one will feel happy or down while eating carrots). Fast-food restaurants predominantly offer highly processed foods to the exclusion of minimally processed foods (Kirkpatrick et al., 2014). Environments that predominantly offer minimally processed foods (e.g., a farmer’s market) might affect minimally processed food expectancies. Overall, the selectivity of the results emphasizes the limitations of an environment to influence expectancies about a behavior more generally. Future research might continue exploring the effect of different environments on food expectancies and the extent to which environments affect expectancies more generally.
Palatable eating coping motives scores did not moderate the effect of the simulated fast-food restaurant on food expectancies. The lack of moderation may underscore the ubiquitous effect of this environment: There was no evidence that it differentially affected food expectancies among individuals according to whether they typically eat highly processed foods to cope with negativity. The lack of moderation also adds to the literature finding inconsistent moderation effects of emotional eating in laboratory mood inductions (Bongers & Jansen, 2016). It is possible that these kinds of moderation effects cannot be detected when self-report methods are used to identify those who eat for emotional reasons (Bongers & Jansen, 2016); however, the PEMS used in this study has been validated against real-time sampling of reasons for eating highly processed foods among college students (Boggiano, Wenger, Turan, Tatum, Sylvester, et al., 2015).
Although palatable eating coping motives scores did not moderate the effect of the simulated fast-food restaurant on food expectancies, greater palatable eating coping motives scores did predict greater positive and negative highly processed food expectancies and greater negative (but not positive) minimally processed food expectancies. Links between palatable eating coping motives scores and food expectancies are important to understand because palatable eating coping motives scores are associated with weight gain among college students (Boggiano, Wenger, Turan, Tatum, Morgan, & Sylvester, 2015). More broadly, individual differences in emotional eating are associated with clinical outcomes such as metabolic syndrome, diabetes, and depression (Epel et al., 2004; Ouwens et al., 2009; Tsenkova et al., 2013). The links observed between palatable eating coping motives scores and positive highly processed food expectancies may reflect the strong preference and tendency to eat highly not minimally processed foods for emotional reasons—a preference conserved across species (Adam & Epel, 2007). In addition, although negative expectancies may lead to less of a behavior, the links between negative expectancies and psychopathology may be more complex (Mann et al., 1987). People with substance use disorder, for example, have experienced negative consequences of substance use (e.g., quit attempt failures, guilt), which could increase their tonic levels of negative substance expectancies (McMahon et al., 1994). Individuals scoring higher in palatable eating coping motives may have therefore endorsed greater negative food expectancies because they have similarly experienced negative consequences from eating certain foods (e.g., diet failure; Deluchi et al., 2017; Nijs & Franken, 2012). 2 Future research might test this directly.
The associations between palatable eating coping motives scores and some types of food expectancies might indicate these constructs are capturing the same latent construct, yet the associations were small in magnitude. In addition, the experimental condition did not affect palatable eating coping motives scores (unlike food expectancies). Although expectancies and motives may be related, expectancy measures require individuals to think about the anticipated consequence of a behavior (e.g., “I expect to feel happy while eating highly processed food”), and motives measures require individuals to think about why they drink (e.g., “I eat highly processed foods to forget my worries”), which makes these constructs conceptually distinct (Kuntsche et al., 2010). For example, one could expect to feel happy while eating highly processed foods but not eat those foods to forget their worries. In the alcohol use literature, researchers have demonstrated that alcohol expectancies and drinking motives were statistically distinct even when nearly identical items were used to assess each construct (e.g., expectancy = “How likely is it that you would be sociable if you drink alcohol?”; motive = “How often have you drunk alcohol to be sociable?”; Kuntsche et al., 2010). Moreover, a large twin study indicated that although drinking motives were genetically heritable, environmental influences shaped alcohol expectancies (Agrawal et al., 2008). Future research in the area of palatable eating motives and food expectancies could empirically test for distinctions in a similar fashion.
In accordance with expectancy theory, increases in positive highly processed food expectancies caused participants to eat a greater number of calories from highly processed foods at the expense of calories from minimally processed foods. These results are theoretically critical because they extend a main tenet of expectancy theory into a novel domain of behavior, and they show that the temporary distortions in expectancies caused by the environment affect environment-relevant behavior. They also have important real-world implications. Diets that largely consist of highly processed foods at the expense of minimally processed foods cause weight gain and predict chronic disease and premature death (Hall et al., 2019; Monteiro et al., 2018; Schnabel et al., 2019). These results suggest that a psychological intervention targeting positive highly processed food expectancies could minimize the impact of a fast-food restaurant on food intake, which may improve diet. Indeed, interventions reducing positive alcohol expectancies consistently decrease how often young adults drink and how much alcohol they consume (Scott-Sheldon et al., 2012), and interventions reducing thinness expectancies have decreased disordered eating (Annus et al., 2008). However, before developing any expectancy-based dietary interventions, interventionists must carefully pilot how to effectively change positive highly processed food expectancies. For instance, the current study data show that expecting to feel negative emotions while eating highly processed foods was not associated with less ad libitum food intake. It thus may be critical to reduce positive highly processed food expectancies without simultaneously increasing negative expectancies in their place (Cummings, Joyner, & Gearhardt, 2020).
These results should be interpreted in light of study limitations. In the current study, there was very limited diversity regarding race/ethnicity and level of parental education among the sample. In prior work, Black compared with non-Black participants rated food expectancies at higher levels, but this was not replicated here, perhaps because of the sample’s limitation (Cummings, Joyner, & Gearhardt, 2020). There are disproportionately more fast-food restaurants in predominantly Black neighborhoods (compared with predominantly White neighborhoods) in the United States (P. James et al., 2014). Compared with non-Black populations, Black people also report greater daily energy intake from highly processed foods (Baraldi et al., 2018) and are more likely to die prematurely from chronic disease (Van Dyke et al., 2018). Future research on the effect of an environment on food expectancies should consider how effects emerge in context of these racial/ethnic disparities and other disparities like those driven by socioeconomic status (Langer et al., 2018).
The current sample also was young and had an average BMI in the “normal” range; this may limit the clinical relevance of the results because people with overweight and obesity are more likely to have experienced clinical outcomes like cardiovascular diseases (Guh et al., 2009). However, body fat doubles from young adult to middle adulthood (Shimokata et al., 1989), and younger compared with older adults report greater intake of highly processed foods (Howarth et al., 2007). Young adults may therefore represent a key population in which to study food expectancies because of the potential clinical impact of dietary interventions targeting food expectancies in this age group. That is, reducing positive highly processed food expectancies among young adults may prevent the development of clinical outcomes later in adulthood. Successful substance use interventions targeting substance expectancies have likewise been delivered to this age group (Scott-Sheldon et al., 2012). Moreover, a diet largely consisting of highly processed foods at the expense of minimally processed foods has been found to be associated with some clinical outcomes independent of BMI (Chen et al., 2020).
The observed effect size for the effect of the simulated fast-food restaurant on positive highly processed food expectancies was small, which is consistent with the one prior study that reported an effect size for the effect of a bar on positive alcohol expectancies (Monk & Heim, 2013). This suggests that environments minimally affect expectancies, at least in this age group, and larger sample sizes may be needed to detect similar effects in the future. Future research might investigate other factors that affect food expectancies such as parental, peer, and media influence, which are implicated in alcohol expectancies (Smit et al., 2018), and future research might determine which factors have larger effects on food expectancies. Nevertheless, even the small observed increase in positive highly processed food expectancies led participants to eat a greater number of highly processed food calories, and very small increases in daily caloric intake (~100 kcal/day) increase body weight over time (Hill et al., 2003). This suggests that repeatedly experiencing small increases in positive highly processed food expectancies could have larger net effects on health, which is important to consider when roughly 35% of U.S. adults visit a fast-food restaurant daily (Nguyen & Powell, 2014). Future research should directly test this.
Although the environmental condition indirectly explained ad libitum food intake through positive highly processed food expectancies, it also directly explained the behavior independent of food expectancies. The lack of full mediation may reflect that the available measurement method for food expectancies is an explicit, self-report questionnaire, which may not be sensitive enough to fully capture environmental effects. A meta-analysis on explicit and implicit (e.g., free associates, implicit association test) measures of alcohol expectancies, for instance, indicated that explicit and implicit measures appear to uniquely capture variance in drinking behavior (Reich et al., 2010). In addition, it is likely that food expectancies represent one of multiple psychological mechanisms through which environments affect food intake; for example, in accordance with incentive-sensitization theory, food “wanting” has mediated environmental effects on food intake (Joyner et al., 2017). To fully explain food intake, researchers might consider simultaneously measuring multiple mediators in future research.
In conclusion, translating principles from expectancy theory into the study of food intake is an emerging research area. The current study provided evidence for the situational-specificity hypothesis by showing that an environment affects environment-relevant food expectancies and that this affects food intake among young adults. In particular, greater positive highly processed food expectancies caused young adults to eat a greater number of calories from highly processed foods at the expense of calories from minimally processed foods. A diet that largely consists of highly processed foods at the expense of minimally processed foods is associated with a number of detrimental clinical outcomes (Chen et al., 2020). Future psychological science on food expectancies will shed light on overlooked social, cognitive, and emotional mechanisms relevant to this pattern of food intake. This kind of research may support the development of dietary interventions that target food expectancies, which may improve diet quality and benefit public health.
Supplemental Material
sj-pdf-1-cpx-10.1177_21677026211004582 – Supplemental material for Extending Expectancy Theory to Food Intake: Effect of a Simulated Fast-Food Restaurant on Highly and Minimally Processed Food Expectancies
Supplemental material, sj-pdf-1-cpx-10.1177_21677026211004582 for Extending Expectancy Theory to Food Intake: Effect of a Simulated Fast-Food Restaurant on Highly and Minimally Processed Food Expectancies by Jenna R. Cummings, Lindzey V. Hoover, Meredith I. Turner, Kalei Glozier, Jessica Zhao and Ashley N. Gearhardt in Clinical Psychological Science
Footnotes
Acknowledgements
We acknowledge Lindsey Parnarouskis, Emma Schiestl, Alexandria Bodfish, Lily Carlson, Aviva Hirsch, Jolie Horne, Benjamin Hsu, Elizabeth Kennedy, Afeefah Khan, Zoe Kolender, Riley Olson, and Rhianna Vergeer for their hard work related to developing the study concept, contributing to the study design, or collecting data.
Transparency
Action Editor: Stefan G. Hofmann
Editor: Kenneth J. Sher
Author Contributions
All of the authors developed the study concept and contributed to the study design. L. V. Hoover, M. I. Turner, K. Glozier, and J. Zhao collected data. J. R. Cummings performed the data analysis and interpretation under the supervision of A. N. Gearhardt. L. V. Hoover confirmed reported estimates based on publicly available data and syntax. K. Glozier conducted an initial literature review. J. R. Cummings drafted the manuscript with assistance from M. I. Turner in drafting the Method section. J. Zhao confirmed the manuscript adhered to American Psychological Association and journal guidelines. All of the authors provided revisions and approved the final manuscript for submission.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
