Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Mar 1.
Published in final edited form as: Psychol Addict Behav. 2019 Dec 12;34(2):403–413. doi: 10.1037/adb0000544

Development and Preliminary Validation of the Anticipated Effects of Food Scale

Jenna R Cummings 1, Michelle A Joyner 2, Ashley N Gearhardt 3
PMCID: PMC7064385  NIHMSID: NIHMS1060516  PMID: 31829666

Abstract

Expectancy Theory has been widely applied in substance use research but has received less attention in eating behavior research. Measuring food expectancies, or the anticipated outcomes of eating specific foods, holds theoretical and practical promise for investigations into nonhomeostatic eating behavior. The current study developed and assessed the psychometric properties of a novel measure of positively and negatively valenced, highly (e.g., sweets, salty snacks, fast foods, sugary drinks) and minimally (e.g., fruits, vegetables) processed food expectancies. The Anticipated Effects of Food Scale (AEFS) was adapted from a self-report of alcohol expectancies, piloted for item generation/retention and readability, and preliminarily validated in an adult sample (N = 247; Mage = 36.84; 53.3% male; 74.5% White). Consistent with the substance expectancies literature, AEFS positive highly processed food expectancies were associated with greater added sugars intake (r = .17, p = .009) and food addiction symptoms (r = .56, p < .001). Of note, AEFS negative minimally processed food expectancies were robustly associated with food addiction symptoms (r = .81, p < .001) and, together with AEFS positive highly processed food expectancies, explained 67% of the variance in food addiction symptoms. Furthermore, AEFS food expectancies demonstrated incremental validity with food addiction symptoms above and beyond general eating expectancies. The AEFS seems to be a psychometrically sound measure and can be used to investigate cognitive-affective mechanisms implicated in highly processed food intake and food addiction. Moreover, the present results provide new insight into potential food expectancy challenge intervention approaches for preventing nonhomeostatic eating behavior.

Keywords: eating behavior, expectancies, food addiction, highly processed food

There is growing interest in utilizing transdiagnostic approaches in the scientific study of substance use and nonhomeostatic eating behavior, or eating in response to cues other than hunger (Racine, Hagan, & Schell, 2019). One theory widely applied in substance use research but that has received less attention in nonhomeostatic eating behavior research is Expectancy Theory (Goldman, Del Boca, & Darkes, 1999). This theory posits that individuals store in memory information about the outcomes of substance use (i.e., expectancies) that they learn from vicarious or personal experience. These expectancies, in turn, affect the likelihood of future substance use (Goldman et al., 1999). For example, an individual may learn from family members, advertisements, and/or personal experience that alcohol makes people feel more sociable and lively, and therefore anticipate that when they drink alcohol they will feel this way (Montes, Witkiewitz, Pearson, & Leventhal, 2019; Morean, Corbin, & Treat, 2012). Elevated positive alcohol expectancies might encourage an individual to drink more often/heavily than they would otherwise (Montes et al., 2019; Morean et al., 2012).

Indeed, a large body of work demonstrates that elevated positive substance expectancies have been linked to substance use. Greater positive alcohol, tobacco cigarette, and marijuana expectancies have been linked to greater alcohol use (Goldman et al., 1999; Morean et al., 2012), tobacco cigarette smoking (Cohen, McCarthy, Brown, & Myers, 2002), and marijuana use (Buckner & Schmidt, 2008), respectively. In adolescents who have yet to try any of these substances, greater positive substance expectancies have predicted the onset of substance use (Montes et al., 2019). In adults, greater positive substance expectancies have predicted the onset and persistence of substance use disorders (Kilbey, Downey, & Breslau, 1998). A smaller body of work shows that elevated negative substance expectancies (e.g., alcohol leads to deteriorated cognitive and behavioral functioning) have correspondingly linked to less substance use (Sharkansky & Finn, 1998; Wiers, Hoogeveen, Sergeant, & Gunning, 2006); however, this link is less consistent with some studies instead showing that greater negative substance expectancies were associated with greater substance use (Mann, Chassin, & Sher, 1987; McMahon, Jones, & O’Donnell, 1994). Overall, the scientific literature has established that substance expectancies are key cognitive factors implicated in substance use and the development of substance use disorders.

Moreover, substance expectancies are modifiable intervention targets relevant to the prevention of problematic substance use and/or substance use disorders. In particular, interventionists have challenged an individual’s positive substance expectancies to reduce these expectancies and thereby reduce substance use (Darkes & Golman, 1993). In the domain of alcohol use, meta-analysis indicates that alcohol expectancy challenge interventions have a small (Cohen’s d = 0.23–0.28) but consistent effect on reducing positive alcohol expectancies, drinking frequency, and heavy drinking levels (Scott-Sheldon, Terry, Carey, Garey, & Carey, 2012). The scientific study of substance expectancies has therefore proven not only to be theoretically important but also practically important.

Given this strong and substantive substance expectancies literature, applying Expectancy Theory to the scientific study of nonhomeostatic eating behavior likely is a prosperous research direction. The Eating Expectancy Inventory (EEI; Hohlstein, Smith, & Atlas, 1998), the existing measure that applies Expectancy Theory in nonhomeostatic eating behavior research, assesses expectancies for reinforcement from eating (food types not specified). Researchers have primarily used this measure to identify cognitive factors implicated in dimensions of nonhomeostatic eating like distress over nonhomeostatic eating (i.e., negative affect arising in response to eating), loss of control over eating (i.e., losing self-regulatory capacity while eating), and overeating (i.e., consuming a large amount of food in a short time period; Racine et al., 2019). For example, researchers have shown with the EEI that expectancies including “eating helps me manage negative affect” and “eating leads to loss of control” were associated with anorexia and/or bulimia nervosa diagnosis, and expectancies that “eating is useful as a reward” were associated with absence of these diagnoses (Hohlstein et al., 1998). However, there are other dimensions of nonhomeostatic eating including emotional eating (i.e., eating in response to emotions) and “food addiction” (i.e., vulnerability to foods that trigger an addictive-like eating response), which are distinct in regard to their associations with clinical outcomes such as Body Mass Index (BMI), depression, and general psychopathology (e.g., emotion dysregulation, negative self-perception; Racine et al., 2019). The EEI may be limited in its applicability to understanding the cognitive factors implicated in all dimensions of nonhomeostatic eating.

Specifically in the context of food addiction, positive and negative “food expectancies,” or the anticipated positively and negatively valenced outcomes of eating certain foods, might provide key insight. A growing body of evidence demonstrates that highly processed foods containing refined carbohydrates and/or fat (e.g., sweets, salty snacks, fast foods, sugary drinks)—in contrast to minimally processed foods (e.g., fruits, vegetables)—may facilitate an addictive-like eating response through greater engagement of neural reward circuitry (DiFeliceantonio et al., 2018), greater subjective experiences of craving and loss of control (Schulte, Avena, & Gearhardt, 2015; Schulte, Smeal, & Gearhardt, 2017), and greater subjective experiences of distress when intake is restricted (Falbe, Thompson, Patel, & Madsen, 2018; Schulte, Smeal, Lewis, & Gearhardt, 2018). Scores on the Yale Food Addiction Scale (YFAS), a measure of food addiction based on the diagnostic criteria for substance use disorders (Gearhardt, Corbin, & Brownell, 2009), have been associated with greater typical highly processed food intake (Lemeshow et al., 2018; Pursey, Collins, Stanwell, & Burrows, 2015) in addition to higher BMI, less treatment success after weight-loss intervention, greater general psychopathology, and greater rates of depression (Meule & Gearhardt, 2014). Identifying expectancies specifically about highly versus minimally processed foods may thus yield vital insight into food addiction and related clinical outcomes.

In sum, there is a critical gap in translating Expectancy Theory from substance use research into nonhomeostatic eating behavior research, and a need for a psychological measure that can capture the level of one’s positive and negative food expectancies. The primary objectives of the current study were to pursue next steps in this promising line of research by: (a) developing and preliminarily validating a novel measure of positive and negative food expectancies that incorporated different food types (highly and minimally processed foods), and (b) identifying food expectancies most strongly implicated in food addiction. The novel measure of food expectancies—the Anticipated Effects of Food Scale (AEFS)—was adapted from the Anticipated Effects of Alcohol Scale (Morean et al., 2012), which is a well validated and widely used measure of positively and negatively valenced alcohol expectancies (see Procedure for full details on the AEFS scale development procedure). To establish reliability and validity of the AEFS, its factor structure and internal consistency were evaluated, and food expectancies scores from the AEFS were compared to conceptually overlapping and distinct constructs (see Measures for justification of the selected constructs). Finally, to identify food expectancies most strongly implicated in food addiction, the relative associations between different food expectancies and food addiction symptoms were investigated.

Methods

Participants

A total of 341 participants were recruited from Amazon’s Mechanical Turk platform. This sample size was based on standard guidelines for conducting exploratory factor analysis plus an additional 10% due to anticipated noncompliance of some participants on the platform (Costello & Osborne, 2005). Participants were paid $1.00 to complete a 30-minute study (Buhrmester, Kwang, & Gosling, 2011). The inclusion criterion was at least 18 years old with no exclusion criteria. Participants were dropped if they incorrectly answered any of the three quality control questions (e.g., “For this item, please mark the response, ‘Seldom,’ to show us you are following directions”; n = 60), completed the questionnaire in less than 10 minutes (n = 17), and/or they reported improbable values for adult height and weight (e.g., BMI = 7.08; n = 17). Visual inspection of a histogram plot revealed that one participant reported an outlier value for BMI (61.99); this participant was excluded from BMI analyses but retained for other analyses.

The final sample comprised 247 participants (53.3% male). On average, participants were 36.84 years old (SD = 11.27, Range = 21–70). The sample was 74.5% White, 15.4% Black, 6.9% Hispanic/Latinx, 4.5% Asian, 2.4% American Indian or Alaskan Native, and 0.8% other (participants could report more than one race/ethnicity). Average BMI was “overweight” at 26.28 (SD = 5.86, Range = 17.71–55.78; 3.1% “underweight,” 42.3% “normal” weight, 36.6% “overweight,” and 18.1% “obese”). Annual household income was as follows: 6.1% reported less than $10000, 38.4% reported $10000–$39999, 30.7% reported $40000–$69999, 15.6% reported $70000–$99999, and 9.4% reported more than $100000.

Procedure

Development of the Anticipated Effects of Food Scale (AEFS).

The development procedure of the Anticipated Effects of Alcohol Scale (Morean et al., 2012), a well validated and widely used alcohol expectancies measure, served as a framework for developing the novel AEFS. Two different prompts for the AEFS were created: one prompt to assess expectancies about the effects of highly processed food and the other prompt to assess expectancies about the effects of minimally processed food. However, to improve readability of the scale, the colloquial term “junk food” was used to describe highly processed food and the colloquial term “healthy food” was used to describe minimally processed food.

To generate items for the scale, a pilot study was conducted in a sample of 101 individuals recruited from Amazon’s Mechanical Turk [52.50% female, Mage(SD) = 35.96 (14.29)]. Participants freely generated responses to prompts asking them to list positive and negative feelings they would expect to have while eating junk and healthy foods. In total, participants provided 159 unique affective words that were potentially related to the anticipatory effects of highly and minimally processed food intake. To assess the extent to which individuals would endorse each of these affective words as highly and minimally processed food expectancies, a second pilot study was conducted in a sample of 98 undergraduate students recruited from the department’s subject pool [typically 57.1% female, Mage(SD) = 18.98 (1.24)]. Participants rated all 159 affective words on a 6-point Likert scale (1 = Not at all, 6 = Very much) in response to two prompts: “How much do you expect to feel the following feelings while eating JUNK foods?” and “How much do you expect to feel the following feelings while eating HEALTHY foods?” Data for each of these pilot studies are publicly available at: https://osf.io/8kspw/.

Researchers have suggested that substance expectancy measures should assess anticipatory substance effects that vary in terms of affective valence and arousal (Morean et al., 2012). Thus, the current study authors made decisions about item retention for the validation study primarily based on coverage of affective space. However, decisions about item retention were also based on linguistic features (e.g., reading level), redundancy between items, descriptive statistics yielded from the second pilot study including statistics on item endorsement frequency and distribution, and inferential statistics yielded from the second pilot study including inter-item correlations (e.g., only one item would be retained where two items were correlated at r > .80). These decisions narrowed the 159 generated affective words down to 31 selected affective words. To improve comprehension, the prompts were adjusted to include examples of highly and minimally processed foods: “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 while eating JUNK food?” and “Imagine that you are eating HEALTHY food (e.g., fruits, vegetables)… how much do you expect to feel the following feelings while eating HEALTHY food?” The 6-point Likert scale was also adjusted to include more anchors (1 = Definitely not, 2 = Probably not, 3 = Possibly, 4 = Probably, 5 = Very probably, 6 = Definitely).

Since research has suggested that negative affect may arise not during but after an eating occasion (Bennett, Greene, & Schwartz-Barcott, 2013), an exploratory version of the AEFS that assessed anticipated delayed effects of eating each food type (effects 15 minutes after eating) was additionally created for the validation study. Fifteen minutes was selected because research has indicated that post-ingestive, satiety-related brain signaling occurs 15–20 minutes after an eating occasion (Thanarajah et al., 2018). Initial review of this exploratory version, however, showed that there was substantial overlap with the original version. That is, the exploratory version yielded scores that correlated at rs > .83 with scores from the original version and yielded scores that demonstrated the same pattern of results with variables of interest. Subsequently, results are only reported for the original version of the AEFS that assessed anticipated effects of highly and minimally processed foods while eating. This selected version of the AEFS is provided in full in Supplemental Materials.

Validation Study.

The University Institutional Review Board approved the research procedure in accordance with the provisions of the World Medical Association Declaration of Helsinki. Participants completed informed consent. Then, in randomized order, participants completed the AEFS and measures of conceptually overlapping and distinct constructs including the modified Yale Food Addiction Scale 2.0, the National Cancer Institute’s Dietary Screener Questionnaire, the Anticipated Effects of Alcohol Scale, and the EEI. All participants next reported demographics including height and weight, and then were compensated for their time.

Measures

Demographics.

Participants were asked to report their age, sex assigned at birth, race/ethnicity, and annual household income.

Convergent validity.

Modified Yale Food Addiction Scale 2.0 (mYFAS 2.0).

Convergent validity of the AEFS was assessed by evaluating associations with mYFAS 2.0 scores because food expectancies were expected to be implicated in food addiction just as substance expectancies are implicated in substance use disorders (Kilbey et al., 1998). The mYFAS 2.0 (Schulte & Gearhardt, 2017) is a 13-item brief version of the Yale Food Addiction Scale 2.0, which measures addictive-like eating responses to highly processed food based on the Diagnostic and Statistical Manual of Mental Disorders 5th edition criteria for substance use disorders. Sample items include, “Eating the same amount of food did not give me as much enjoyment as it used to,” and “I ate to the point where I felt physically ill.” Participants rated each item on an 8-point Likert scale (1 = Never to 8 = Every day), and whether or not this rating meets the “diagnostic” threshold for each symptom is determined. All symptom values were summed to create a dimensional score of food addiction (M = 2.56, SD = 3.67, Range = 0–11, α = .94).

National Cancer Institute’s Dietary Screener Questionnaire.

Convergent validity of the AEFS was additionally assessed by evaluating associations with added sugars intake, an indicator of highly processed food intake (Monteiro et al., 2018), because food expectancies were expected to be linked to food intake just as substance expectancies are linked to substance use (Buckner & Schmidt, 2008; Cohen et al., 2002; Goldman et al., 1999; Morean et al., 2012). Added sugars intake was selected because of strong evidence linking it to addictive behavior (Ahmed, Guillen, & Vandaele, 2013; Avena, Rada, & Hoevel, 2008; Burger, 2017; Colantuoni et al., 2012) and to clinical outcomes including greater body weight (Morenga, Mallard, & Mann, 2013), cardiovascular disease (Yang et al., 2014), and cancer (Tasevska et al., 2012). The National Cancer Institute’s Dietary Screener Questionnaire (Thompson, Midthune, Kahle, & Dodd, 2017) is a 26-item questionnaire which asks about the frequency of intake in the past month of selected foods and drinks; the current study administered the 9 items needed to assess added sugars intake (i.e., soda, fruit drinks, cookies/cake/pie, doughnuts, ice cream, sugar/honey in coffee/tea, candy, cereal and cereal type). Publicly available scoring algorithms that couple the item frequency responses with sex- and age-specific portion size information were used to generate estimated portion intake of added sugars per day (M = 8.48 tsp./day, SD = 13.06 tsp./day, Range = 0–92.19 tsp./day; Thompson et al., 2017).

BMI.

Convergent validity of the AEFS was also assessed by evaluating associations with BMI because greater intake of highly processed foods can cause weight gain (Hall et al., 2019) and because food addiction is linked with higher BMI (Meule & Gearhardt, 2014); thus, it is plausible that food expectancies are implicated in BMI. Participants were asked to report their height and weight. BMI was calculated using the standard formula (kg/m2).

Discriminant validity.

Anticipated Effects of Alcohol Scale (AEAS).

Discriminant validity of the AEFS was assessed by evaluating associations with alcohol expectancies because it was speculated that information stored in memory about certain foods would be distinct from information stored in memory about alcohol. Indeed, prior work using the EEI demonstrated that “eating helps me manage negative affect” expectancies were not significantly associated with “alcohol aids in tension reduction” expectancies (Fischer, Anderson, & Smith, 2004). The AEAS (Morean et al., 2012) is a 44-item questionnaire that measures alcohol expectancies. Participants were instructed to imagine that they were drinking four (female version) or five (male version) drinks within a two-hour period, and then rate how they expected to feel immediately after drinking and an hour and a half after finishing their last drink. Thus, the scale assessed both the anticipated ascending (e.g., stimulatory) and descending limb (e.g., sedative) effects of alcohol. Participants rated each item on an 11-point Likert scale (0 = Not at all to 10 = Extremely).

For the current study, only items assessing anticipated ascending limb effects of alcohol were scored to parallel the AEFS, which only assessed anticipated effects of food while eating. The Positive-High Arousal-Ascending Limb (i.e., lively, sociable, carefree, fun, attractive, funny; α = .91) and the Positive-Low Arousal-Ascending Limb (i.e., mellow, calm, relaxed; α = .78) subscales were scored by taking the average across their respective items, and a positive alcohol expectancies score was then created by taking the average across these two subscales (M = 7.10, SD = 2.11, Range = 1–11). The Negative-High Arousal-Ascending Limb (i.e., demanding, moody, rude, aggressive, anxious; α = .92) and Negative-Low Arousal-Ascending Limb (i.e., wobbly, woozy, dizzy, ill; α = .89) subscales were scored by taking the average across their respective items, and a negative alcohol expectancies score was then created by taking the average across these two subscales (M = 5.02, SD = 2.50, Range = 1–11).

Incremental validity.

Eating Expectancy Inventory (EEI).

Incremental validity was assessed by evaluating whether scores from the AEFS were associated with YFAS scores over and above scores from the EEI. This would demonstrate whether measuring food expectancies, rather than only expectancies about eating generally, is a potentially clinically meaningful endeavor. The EEI (Hohlstein et al., 1998) is a 34-item questionnaire that measures expectancies for reinforcement from eating and from dieting. Participants rated each item on a 7-point Likert scale (1 = Completely disagree to 7 = Completely agree). The EEI Manage Negative Affect (e.g., When I am feeling depressed or upset, eating can help me take my mind off my problems; M = 3.67, SD = 1.54, Range = 1–6.82, α = .96), the EEI Useful as a Reward (e.g., When I do something good, eating is a way to reward myself; M = 4.68, SD = 1.18, Range = 1–7, α = .74), and the EEI Loss of Control (e.g., When I eat I often feel that I am not in charge of my life; M = 3.26, SD = 1.64, Range = 1–7, α = .83) subscales were scored by taking the average across their respective items.

Data Analytic Plan

Data and syntax are publicly available at: https://osf.io/8kspw/. All variables of interest were reviewed for normality, outliers, and missing data. Less than 1.5% of data were missing; thus, a pairwise deletion approach was utilized. To achieve the primary objective of the current study, exploratory factor analysis (EFA) was conducted to assess the factor structure of the AEFS. A maximum likelihood approach was selected because items were generally normally distributed, and oblique rotation was selected because factors would likely correlate with one another (Costello & Osborne, 2005). Internal consistency (Cronbach’s α) among items was evaluated for the resulting factors. In addition, bivariate correlations were conducted between resulting factors and demographic variables, and any significant demographic correlate was tested as a potential covariate in subsequent regression models. Inclusion of significant demographic correlates did not change the pattern of results in these models; thus, the more parsimonious models are presented here. For estimates from the adjusted models, please see Tables S1 and S2 in Supplemental Materials.

To assess convergent validity, bivariate correlations were conducted between resulting factors and mYFAS 2.0 scores, added sugars intake, and BMI. To assess discriminant validity, bivariate correlations were conducted between resulting factors and positive and negative alcohol expectancies. To assess incremental validity, stepwise linear regression was conducted to test if the resulting factors were associated with mYFAS 2.0 scores over and above eating expectancies measured by conceptually similar subscales from the EEI. To identify which food expectancies were most strongly implicated in food addiction, resulting factors were entered into a simultaneous linear regression to evaluate relative associations with mYFAS 2.0 scores and estimate variance accounted for (ΔR2). All analyses were conducted in SPSS Version 24.

Results

AEFS Factor Structure, Descriptives, & Internal Consistency

EFA of the highly processed food items of the AEFS yielded a scree plot with three factors above the “break,” each with an eigenvalue > 1. EFA of the minimally processed food items of the AEFS yielded a scree plot with two factors above the “break,” both with an eigenvalue > 1. Thus, in accordance with EFA best practices, one-factor, two-factor, three-factor, and four-factor solutions were further investigated (Costello & Osborne, 2005). Number of items in factors and factor loadings across solutions were compared and, for both the highly processed food items and the minimally processed food items, two-factor solutions were selected (labeled “positive” and “negative” food expectancies). These solutions were selected because no factors consisted of less than three items, there were appropriate magnitudes of factor loadings (all factor loadings > .54), and no cross-loadings (i.e., no items with factor loadings > .30 for both factors; Costello & Osborne, 2005). In particular, a three-factor solution for the highly processed food items yielded a third factor including only one item with a loading > .32 (“lazy”; loading = .34), and this item cross-loaded with another factor (loading = .78).

Positive and negative food expectancies were scored by taking the average across their respective highly and minimally processed food items.1 Positive highly processed food expectancies (M = 3.32, SD = 1.19, Range = 1.00–5.80, α = .94), negative highly processed food expectancies (M = 2.71, SD = 1.34, Range = 1.00–6.00, α = .96), positive minimally processed food expectancies (M = 3.96, SD = 1.16, Range = 1.00–6.00, α = .95), and negative minimally processed food expectancies (M = 2.12, SD = 1.40, Range = 1.00–5.67, α = .98) all yielded excellent internal consistency. Table 1 presents means, standard deviations, and factor loadings for each item across the AEFS.

Table 1.

Items & Factor Loadings for the Anticipated Effects of Food Scale

Highly Processed Food Minimally Processed Food
Factor 1 Factor 2 Factor 1 Factor 2
Mean SD Loading Loading Mean SD Loading Loading
+ − + −
Factor 1 + Relaxed 3.58 1.55 .82 −.12 3.93 1.54 .80 −.02
Cheerful 3.40 1.52 .82 −.09 4.06 1.46 .86 −.01
Refreshed 2.97 1.63 .81 .07 4.15 1.47 .86 −.05
Glad 3.41 1.57 .79 −.07 4.26 1.41 .84 −.07
Excited 2.90 1.64 .76 .13 3.15 1.74 .62 .31
Focused 2.96 1.59 .76 .12 4.03 1.57 .72 .03
Calm 3.72 1.41 .75 −.17 4.15 1.38 .74 −.06
Happy 3.97 1.44 .73 −.17 4.33 1.37 .85 −.11
Energized 2.99 1.58 .70 .05 4.16 1.43 .77 −.14
Proud 2.59 1.70 .70 .22 4.21 1.53 .68 .02
Relieved 2.97 1.63 .66 .26 3.75 1.65 .71 .15
Alert 3.25 1.62 .60 .16 3.78 1.63 .62 .08
Soothed 3.52 1.55 .59 .11 3.50 1.60 .71 .12
Comforted 3.73 1.50 .55 .02 3.63 1.54 .74 .09
Content 3.79 1.47 .55 −.15 4.17 1.47 .74 −.13
Factor 2 − Down 2.70 1.65 −.01 .89 2.01 1.44 −.05 .90
Depressed 2.69 1.67 −.05 .87 2.17 1.58 −.07 .88
Disgusting 2.62 1.71 −.07 .83 1.92 1.51 .02 .88
Worried 2.53 1.61 .13 .83 2.00 1.50 .06 .90
Frustrated 2.64 1.68 −.04 .82 2.17 1.64 −.03 .91
Regretful 3.20 1.73 −.16 .80 2.04 1.60 .05 .91
Irritable 2.40 1.51 .04 .79 2.17 1.56 −.05 .85
Anxious 2.55 1.66 .20 .78 2.13 1.60 .08 .89
Ashamed 3.02 1.74 −.12 .77 1.89 1.58 .05 .84
Sluggish 3.16 1.61 −.11 .77 2.07 1.53 .03 .91
Tired 2.80 1.57 .02 .76 2.13 1.54 −.02 .92
Afraid 2.13 1.53 .22 .75 1.89 1.41 .05 .88
Numb 2.60 1.73 .26 .73 2.09 1.62 .03 .92
Lazy 3.02 1.70 −.20 .71 2.08 1.60 .03 .86
Deprived 2.32 1.61 .25 .69 2.35 1.56 −.05 .77
Bored 2.76 1.62 .10 .68 2.50 1.60 −.08 .79
Open in a new tab

Notes: + indicates positive expectancies and − indicates negative expectancies

Demographic Correlates

There were small correlations between age and positive highly processed food expectancies (r = −.16, p = .014), negative highly processed food expectancies (r = −.25, p < .001), and negative minimally processed food expectancies (r = −.28, p < .001) such that younger individuals were more likely to endorse these expectancies. There was a small correlation between sex (dummy code: 0 = men, 1 = women) and positive highly processed food expectancies (r = −.20, p = .002) such that male participants [M(SD) = 3.54(1.16)] were more likely to endorse these expectancies compared to female participants [M(SD) = 3.08(1.18)]. There was also a small correlation between sex and negative minimally processed food expectancies (r = −.21, p = .001) such that male participants [M(SD) = 2.39(1.55)] were more likely to endorse these expectancies compared to female participants [M(SD) = 1.81(1.15)].

In addition, there was a small correlation between race/ethnicity (dummy code: 0 = non-Black, 1 = Black2) and positive highly processed food expectancies (r = .16, p = .013) such that Black participants [M(SD) = 3.76(1.27)] were more likely to endorse these expectancies compared to non-Black participants [M(SD) = 3.24(1.16)]. There was also a small correlation between race/ethnicity and negative highly processed expectancies (r = .14, p = .033) such that Black participants [M(SD) = 3.14(1.49)] were more likely to endorse these expectancies compared to non-Black participants [M(SD) = 2.64(1.30)]. Likewise, there was a small correlation between race/ethnicity and positive minimally processed food expectancies (r = .13, p = .047) such that Black participants [M(SD) = 4.30(1.29)] were more likely to endorse these expectancies compared to non-Black participants [M(SD) = 3.90(1.13)]. There was also a small correlation between race/ethnicity and negative minimally processed food expectancies (r = .19, p = .002) such that Black participants [M(SD) = 2.75(1.75)] were more likely to endorse these expectancies compared to non-Black participants [M(SD) = 2.00(1.30)]. There were no correlations between annual household income and any of the measured food expectancies.

Convergent Validity

Table 2 presents correlation coefficients and statistical significance for associations between AEFS food expectancies and mYFAS 2.0 scores, added sugars intake, and BMI. Greater positive and negative highly processed food expectancies, and greater negative minimally processed food expectancies, showed large correlations with greater mYFAS 2.0 scores, small correlations with greater added sugars intake, and no correlation with BMI. In contrast, greater positive minimally processed food expectancies had a small correlation with mYFAS 2.0 scores, had no correlation with added sugars intake, and had a small correlation with lower BMI.

Table 2.

Correlations between the Anticipated Effects of Food Scale & Food Addiction Symptoms, Added Sugars Intake, and BMI

1. 2. 3. 4. 5. 6. 7.
1. +Highly Processed Food .32*** .41*** .64*** .56*** .17** −.07
2. −Highly Processed Food .42*** .75*** .63*** .19** .04
3. +Minimally Processed Food .29*** .27*** −.03 −.16*
4. −Minimally Processed Food .81*** .26*** −.06
5. Food Addiction Symptoms .29*** .01
6. Added Sugars Intake .05
7. BMI
Open in a new tab

Notes: + indicates positive expectancies and − indicates negative expectancies.

*

p < .05,

**

p < .01,

***

p < .001.

Discriminant Validity

AEFS food expectancies demonstrated small to medium correlations (rs = .21–.42, ps < .001) with greater positive alcohol expectancies, and demonstrated medium to large correlations (rs = .30–.60, ps < .001) with negative alcohol expectancies. Positive highly processed food expectancies showed stronger correlation with positive (compared to negative) alcohol expectancies (r = .42, p < .001). Negative highly processed food expectancies, positive minimally processed food expectancies, and negative minimally processed food expectancies showed stronger correlations with negative (compared to positive) alcohol expectancies (rs = .56, .30, and .60, respectively; all ps < .001).

Incremental Validity

We identified the EEI Manage Negative Affect and EEI Useful as a Reward subscales as conceptually similar to positive highly and minimally processed food expectancies, and we identified the EEI Loss of Control subscale as conceptually similar to negative highly and minimally processed food expectancies. Table 3 presents estimates from the stepwise regression models testing incremental validity of the AEFS relative to these respective subscales. Results indicated that greater positive highly processed food expectancies had a medium association with mYFAS 2.0 scores above and beyond relevant EEI subscales, accounting for an additional 9% of the variance. Positive minimally processed food expectancies had a small association with mYFAS 2.0 scores, accounting for an additional 2% of the variance. In addition, results indicated that negative highly processed food expectancies had a medium association with mYFAS 2.0 scores above and beyond the relevant EEI subscale, accounting for an additional 9% of the variance. Negative highly processed food expectancies had a large association with mYFAS 2.0 scores, accounting for an additional 32% of the variance.

Table 3.

Incremental Validity of the Anticipated Effects of Food Scale

95% Confidence Intervals
R2 ΔR2 B SE β p Lower Upper
Stepwise Model 1
Step 1 .45 .45
 EEI Manage Negative Affect 1.60 0.12 0.67 <.001 1.36 1.83
 EEI Useful as a Reward −1.05 0.15 −0.34 <.001 −1.35 −0.75
Step 2 .54 .09
 +Junk Food 1.08 0.16 0.34 <.001 0.76 1.40
Stepwise Model 2
Step 1 .45 .45
 EEI Manage Negative Affect 1.60 0.12 0.67 <.001 1.36 1.83
 EEI Useful as a Reward −1.05 0.15 −0.34 <.001 −1.35 −0.75
Step 2 .48 .02
 +Healthy Food 0.46 0.15 0.15 .003 0.16 0.77
Stepwise Model 3
Step 1 .39 .39
 EEI Loss of Control 1.40 0.12 0.62 <.001 1.17 1.62
Step 2 .47 .09
 −Junk Food 1.08 0.17 0.39 <.001 0.74 1.42
Stepwise Model 4
Step 1 .39 .39
 EEI Loss of Control 1.40 0.12 0.62 <.001 1.17 1.62
Step 2 .71 .32
 −Healthy Food 1.87 0.12 0.68 <.001 1.64 2.10
Open in a new tab

Notes: + indicates positive expectancies and − indicates negative expectancies.

Food Expectancies & Food Addiction

Table 4 presents estimates from the simultaneous regression model testing the relative associations of different AEFS food expectancies with mYFAS 2.0 scores. By and large, negative minimally processed food expectancies showed the strongest (and a large) association with mYFAS 2.0 scores, yet positive highly processed food expectancies also had a small association with mYFAS 2.0 scores. The total model accounted for 68% of the variance in mYFAS 2.0 scores.

Table 4.

Relative Associations between the Anticipated Effects of Food Scale & Food Addiction Symptoms

95% Confidence Intervals
R2 ΔR2 B SE β p Lower Upper
Simultaneous Model .68 .68
 +Junk Food 0.41 0.18 0.13 .020 0.07 0.76
 −Junk Food 0.34 0.18 0.12 .058 −0.01 0.69
 +Healthy Food 0.00 0.14 0.00 .997 −0.28 0.28
 −Healthy Food 1.80 0.20 0.65 <.001 1.39 2.19
Open in a new tab

Notes: + indicates positive expectancies and − indicates negative expectancies.

Discussion

The current study developed and preliminarily validated a novel measure of food expectancies, the Anticipated Effects of Food Scale (AEFS), in an adult sample. The AEFS measures the positively and negatively valenced anticipated outcomes of eating different types of food including highly (e.g., sweets, salty snacks, fast foods, sugary drinks) and minimally processed foods (e.g., fruits, vegetables). The development and preliminary validation of the AEFS is important in light of growing evidence documenting that intake of highly processed foods containing refined carbohydrates and/or fat may facilitate addictive-like eating responses (DiFeliceantonio et al., 2018; Falbe et al., 2018; Schulte et al., 2015; Schulte et al., 2017; Schulte et al., 2018). Also, borrowing Expectancy Theory from substance use research (Goldman et al., 1999; Morean et al., 2012) and applying it in the context of nonhomeostatic eating behavior research will likely yield a number of promising avenues for future investigation.

The Anticipated Effects of Alcohol Scale (AEAS; Morean et al., 2012) served as a framework for the development of the AEFS, which included piloting for item generation/retention and readability. Results indicated that the AEFS had a two-factor structure (i.e., positive, negative), and demonstrated strong internal consistency as well as good convergent and incremental validity. The two-factor structure of the AEFS is different from the AEAS, which has a four-factor structure (i.e., positive high-arousal, positive low-arousal, negative high-arousal, negative low-arousal). The difference in factor structures between the AEFS and the AEAS likely reflects that alcohol causes distinct biphasic effects that differ in arousal level (i.e., stimulatory and sedative) but food does not. Thus, participants may have rated food expectancy items more similarly based on valence versus arousal. However, future validation studies of the AEFS should replicate its factor structure.

Results indicated that greater positive highly processed food expectancies (i.e., anticipating that one will feel “cheerful,” “relaxed,” etc. from eating sweets, salty snacks, fast foods, sugary drinks) were associated with greater food addiction symptoms and added sugars intake. This is very consistent with the substance use literature, which shows that greater positive substance expectancies have been associated with greater substance use (Buckner & Schmidt, 2008; Cohen et al., 2002; Goldman et al., 1999; Morean et al., 2012) and the likelihood of developing a substance use disorder (Kilbey et al., 1998). Greater negative highly processed food expectancies (i.e., anticipating that one will feel “down,” “depressed,” etc. from eating sweets, salty snacks, fast foods, sugary drinks) were also associated with greater food addiction symptoms (albeit at trend levels when accounting for other food expectancies) and added sugars intake. In the substance use literature, some studies have likewise shown that greater negative substance expectancies were associated with greater substance use (Mann et al., 1987; McMahon et al., 1994)—albeit it is generally theorized that greater negative substance expectancies would encourage less substance use (Montes et al., 2019; Morean et al., 2012). One proposed explanation is that the association between negative substance expectancies and substance use varies as a function of an individual’s risk for substance use disorder (Mann et al., 1987). For example, for those who show low risk for alcohol use disorder (e.g., those who drink heavily at a few social occasions), positive alcohol expectancies alone may predict greater alcohol use; for those who show high risk for alcohol use disorder, both positive and negative alcohol expectancies may predict greater alcohol use (Mann et al., 1987). This may be because those who show high risk versus low risk for alcohol use disorder have experienced negative consequences of alcohol use and/or have been motivated to quit or restrain alcohol use, which elevates their negative alcohol expectancies (McMahon et al., 1994).

Individuals who show high risk for food addiction may therefore anticipate positive affect from highly processed foods while simultaneously anticipating negative affect3 because of experiences with negative consequences of highly processed food intake (e.g., weight gain), and/or because of motivation to quit or restrain intake. Indeed, individuals engaging in nonhomeostatic eating behaviors often show ambivalent approach-avoidance patterns towards highly processed foods (Deluchi, Costa, Friedman, Goncalves, & Bizarro, 2017; Nijs & Franken, 2012). Also, continuing to use a substance despite awareness of the negative consequences is a hallmark feature of substance use disorders (Leshner, 1997).

Greater positive minimally processed food expectancies (i.e., anticipating that one will feel “cheerful,” “relaxed,” etc. from eating fruits, vegetables) were weakly associated with food addiction symptoms (and not associated when accounting for other food expectancies) and were not associated with added sugars intake. This corroborates with prior work wherein individuals with food addiction symptoms report having far fewer problems with eating minimally versus highly processed foods (Schulte et al., 2015). Greater negative minimally processed food expectancies (i.e., anticipating that one will feel “down,” “depressed,” etc. from eating fruits, vegetables), however, were associated with greater added sugars intake and very strongly associated with greater food addiction symptoms. Furthermore, relative to the three other food expectancies measured, greater negative minimally processed food expectancies demonstrated the most robust association with food addiction symptoms.

Why might negative minimally processed food expectancies be so strongly implicated in food addiction? Non-human animal models demonstrate that extended access to highly processed foods (i.e., bacon, cheesecake, frosting) reduces neural dopamine receptor sensitivity, which causes compulsive drive for highly processed foods but not other foods (i.e., chow)—even when these other foods are far more easily accessible (see DiFeliceantonio & Small, 2018 for a review on the non-human animal research and the preliminary translational work). In accordance with this framework, and given that individuals with food addiction tend to eat high amounts of highly processed foods (Lemeshow et al., 2018; Pursey et al., 2015), it is plausible that neural reward processing adaptations have led to lower preference for minimally processed foods among these individuals. Individuals with food addiction may therefore strongly anticipate that minimally processed foods will cause negative affective experiences. Indeed, a recent study found that, for individuals with food addiction compared to weight-matched controls, images of minimally processed foods caused robust reductions in neural activation patterns that have been associated with cue-induced craving (Schulte, Yokum, Jahn, & Gearhardt, 2019).

These results offer practical implications. When interventionists have targeted substance expectancies to prevent substance use and/or substance use disorders they have generally focused on reducing positive substance expectancies (Darkes & Golman, 1993; Scott-Sheldon et al., 2012). To prevent highly processed food intake and/or food addiction, interventionists may similarly want to focus on reducing positive highly processed food expectancies. Yet, interventionists should note that—because negative highly processed food expectancies were associated with highly processed food intake and food addiction symptoms (albeit at trend levels when accounting for other food expectancies)—it may be critical to reduce positive highly processed food expectancies without simultaneously increasing negative highly processed food expectancies. In addition, the current study results suggest that it may be most beneficial to reduce negative minimally processed food expectancies in addition to reducing positive highly processed food expectancies (together these expectancies accounted for 67% of the variance in food addiction symptoms). This dual approach might confer additional health benefits because greater intake of minimally processed foods reduces risk for cardiometabolic disease (Cavallo, Horino, & McCarthy, 2016). Future research might consider using substance expectancy challenge interventions as guides for food expectancy challenge interventions while taking into account the distinctive findings regarding food expectancies.

Although greater positive and negative highly processed food expectancies as well as negative minimally processed food expectancies were strongly linked with added sugars intake and food addiction symptoms, they were not associated with BMI. These results were unexpected but they underscore that, while sometimes related, food addiction and BMI are distinct clinical outcomes (Meule & Gearhardt, 2014). BMI is heterogeneously determined by several factors including diet, exercise, psychological stress, and genetics (Borecki et al., 1998). It is possible that not accounting for the aforementioned factors made it difficult to detect whether some food expectancies were associated with BMI.

On the other hand, greater positive minimally processed food expectancies were associated with lower BMI. This could reflect that they were the only food expectancies not associated with added sugars intake, and highly processed foods containing added sugars can cause weight gain (Hall et al., 2019). Moreover, those who anticipate positive outcomes from eating minimally processed food expectancies may eat more fruits and vegetables, which has been shown to prevent weight gain (Mozaffarian, Hao, Rimm, Willett, & Hu, 2011). The unique association between positive minimally processed food expectancies and BMI may offer practical implications to interventionists targeting obesity. That is, increasing the likelihood that individuals anticipate positive affective experiences while eating minimally processed foods may be beneficial for preventing obesity. Some have likewise suggested interventionists should encourage individuals to focus on pleasurable aspects of eating minimally processed foods rather than on health benefits (Pettigrew, 2016). This could be especially helpful in childhood (Marty, Chambaron, Nicklaus, & Monnery-Patris, 2018), a sensitive period for preventing weight gain to promote lifelong health (Wing & Phelan, 2005).

In regards to discriminant validity of the AEFS, results unexpectedly indicated that food and alcohol expectancies were associated; this is inconsistent with prior work showing that “eating helps me manage negative affect” expectancies were not significantly associated with “alcohol aids in tension reduction” expectancies (albeit they were correlated at r = .13; Fischer et al., 2004). The reason for these results is unclear, yet some researchers have suggested that shared neural reward mechanisms or co-use of different substances might promote overlapping substance expectancies (Walther, Pedersen, Gnagy, Pelham, & Molina, 2019). Very few studies have simultaneously investigated different substance expectancies but there is some evidence that alcohol and cigarette expectancies are associated (McCarthy & Thompsen, 2006), and that alcohol and marijuana expectancies are associated (Walther et al., 2019). However, these studies generally find substance-specific associations between substance expectancies and use (e.g., positive alcohol expectancies predict alcohol but not marijuana use; McCarthy & Thompsen, 2006; Walther et al., 2019). Future research might therefore consider investigating whether food expectancies are associated with substance use (e.g., drinking behavior) to better establish discriminant validity of the AEFS.

Incremental validity of the AEFS was demonstrated by showing that highly and minimally processed food expectancies accounted for considerable variance in food addiction symptoms over and above relevant subscales from the EEI. This established that measuring different food expectancies—rather than only expectancies about eating generally—is a potentially clinically meaningful endeavor especially relevant to food addiction. Researchers might consider including the EEI and/or the AEFS in future studies dependent upon which dimensions of nonhomeostatic eating and clinical outcomes they are most interested in studying. It is possible that positive and negative highly and minimally processed food expectancies may be relevant to the study of other dimensions beyond food addiction. For example, when individuals engage in emotional eating, they are often consuming highly rather than minimally processed foods (Adam & Epel, 2007). Likewise, researchers might consider only administering portions of the AEFS when practically constrained.

Results should be interpreted in light of study limitations. The causal role of food expectancies cannot be established from the current study. In the broader substance use literature, substance expectancies have been experimentally and prospectively shown to impact substance use (Goldman et al., 1999; Kilbey et al., 1998; Montes et al., 2019). This precedent provides confidence that food expectancies could serve a causal role in highly processed food intake and food addiction, but it will be important for future research to establish this with experiments and longitudinal work. In addition, the current study did not examine test-retest reliability of the AEFS, and future research establishing stability of AEFS scores in the short-term (e.g., over a two week period) would bolster confidence in the scale’s reliability and be consistent with findings from the substance expectancies literature (Morean et al., 2012). The current study also used a self-reported dietary intake measure. The sex- and age-specific dietary calculations from this measure have been well validated (Thompson et al., 2017). However, future research should test whether AEFS food expectancies would be associated with objectively measured highly processed food intake. In doing so, the operationalization of highly processed food intake could be expanded beyond added sugars intake and could include, for example, intake of foods high in refined carbohydrates and fat (e.g., pizza, cookies). Lastly, future research might consider measuring food expectancies in more diverse populations. This could be especially important given that Black participants endorsed greater food expectancies, and a large U.S. disparity is the substantially higher prevalence of obesity among Black versus White populations (Hales et al., 2018; Ogden et al., 2012).

Moreover, although the current study provides preliminary guidance on how scores on the AEFS can be clinically informative of food addiction risk, more research will shed light on the clinical utility of the AEFS. For instance, future research using a larger, representative sample might test associations of the four identified food expectancies with food addiction symptoms by subgroups of the population (e.g., by gender, by race). This might help identify whether certain food expectancies are more strongly associated with food addiction symptoms for certain groups of individuals. Additionally, future research might test associations of the four identified food expectancies with food addiction symptoms across developmental stages. If negative highly processed food expectancies emerge in response to experiencing negative consequences of highly processed food intake and/or being motivated to quit or restrain intake, it would be expected that they may elevate at stages when negative consequences of highly processed food intake are salient (e.g., after weight gain). This kind of research may help refine Expectancy Theory as it applies to food addiction. It also may inform on the discrepancy between Expectancy Theory and empirical findings that greater negative substance expectancies were associated with greater substance use (Mann et al., 1987; McMahon et al., 1994).

Limitations of the current study notwithstanding, the novel AEFS seems to be a psychometrically sound measure of food expectancies, and it demonstrated excellent internal consistency and good convergent and incremental validity in an adult sample. Importantly, the AEFS will facilitate future research investigating cognitive-affective mechanisms implicated in food addiction. The AEFS in combination with the rich substance expectancies literature (Goldman et al., 1999) leaves investigators well positioned to further apply Expectancy Theory in nonhomeostatic eating behavior research. This transdiagnostic approach will likely be theoretically and practically important for the field.

Supplementary Material

tables

Acknowledgments

Jenna R. Cummings was supported by Award Number T32HD079350 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development. The present research was supported by funds allocated for research through T32HD079350. The content of the manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the Eunice Kennedy Shriver National Institute of Child Health and Human Development.

Footnotes

The data presented in this manuscript have not been disseminated previously (e.g., presented at a conference). The data have been shared on the Open Science Framework at https://osf.io/8kspw/.

1

Due to research error, the tense of one item of the AEFS was presented inconsistently (i.e., either as “disgusting” or “disgusted”). We re-scored the AEFS without this item and re-ran all analyses. Results did not differ whether or not this item was included. Below, we label the item as “disgusting.”

2

Race/ethnicity was dummy coded as 0 = non-Black and 1 = Black for the current study because the prevalence of obesity is substantially higher in Black compared to White populations in the U.S. (Hales et al., 2018; Ogden, Carroll, Kit, & Flegal, 2012). Thus, identifying differences in food expectancies across non-Black and Black populations may point to potential cognitive-affective mechanisms relevant to this disparity in obesity prevalence.

3

In post-hoc analysis, an interaction between positive and negative highly processed food expectancies was associated with food addiction symptoms; those high in both types of highly processed food expectancies showed greater food addiction symptoms. Please see Table S3 in Supplemental Materials for these results.

Contributor Information

Jenna R. Cummings, University of Michigan, Ann Arbor

Michelle A. Joyner, University of Michigan, Ann Arbor

Ashley N. Gearhardt, University of Michigan, Ann Arbor

References

  1. Adam TC, & Epel ES (2007). Stress, eating and the reward system. Physiology & Behavior, 91(4), 449–458. [DOI] [PubMed] [Google Scholar]
  2. Ahmed SH, Guillem K, & Vandaele Y (2013). Sugar addiction: pushing the drug-sugar analogy to the limit. Current Opinion in Clinical Nutrition & Metabolic Care, 16(4), 434–439. [DOI] [PubMed] [Google Scholar]
  3. Avena NM, Rada P, & Hoebel BG (2008). Evidence for sugar addiction: behavioral and neurochemical effects of intermittent, excessive sugar intake. Neuroscience & Biobehavioral Reviews, 32(1), 20–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bennett J, Greene G, & Schwartz-Barcott D (2013). Perceptions of emotional eating behavior. A qualitative study of college students. Appetite, 60(1), 187–192. doi: 10.1016/j.appet.2012.09.023 [DOI] [PubMed] [Google Scholar]
  5. Borecki IB, Higgins M, Schreiner PJ, Arnett DK, Mayer-Davis E, Hunt SC, & Province MA (1998). Evidence for multiple determinants of the Body Mass Index: The National Heart, Lung, and Blood Institute Family Heart Study. Obesity Research, 6(2), 107–114. [DOI] [PubMed] [Google Scholar]
  6. Buckner JD, & Schmidt NB (2008). Marijuana effect expectancies: Relations to social anxiety and marijana use problems. Addictive Behaviors, 33(11), 1477–1483. doi: 10.1016/j.addbeh.2008.06.017 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Buhrmester M, Kwang T, & Gosling SD (2011). Amazon’s Mechanical Turk: A New Source of Inexpensive, Yet High-Quality, Data? Perspectives on Psychological Science, 6(1), 3–5. doi: 10.1177/1745691610393980 [DOI] [PubMed] [Google Scholar]
  8. Burger KS (2017). Frontostriatal and behavioral adaptations to daily sugar-sweetened beverage intake: a randomized controlled trial. The American Journal of Clinical Nutrition, 105(3), 555–563. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Cavallo DN, Horino M, & McCarthy WJ (2016). Adult intake of minimally processed fruits and vegetables: Associations with cardiometabolic disease risk factors. Journal of the Academy of Nutrition & Dietetics, 116(9), 1387–1394. doi: 10.1016/j.jand.2016.03.019 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Colantuoni C, Rada P, McCarthy J, Patten C, Avena NM, Chadeayne A, & Hoebel BG (2002). Evidence that intermittent, excessive sugar intake causes endogenous opioid dependence. Obesity Research, 10(6), 478–488. [DOI] [PubMed] [Google Scholar]
  11. Cohen LM, McCarthy DM, Brown SA, & Myers MG (2002). Negative affect combines with smoking outcome expectancies to predict smoking behavior over time. Psychology of Addictive Behaviors, 16(2), 91–97. doi: 10.1037/0893-164X.16.2.91 [DOI] [PubMed] [Google Scholar]
  12. Costello AB, & Osborne JW (2005). Best practicies in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment Research & Evaluation, 10(7), 1–9. [Google Scholar]
  13. Darkes J, & Golman MS (1993). Expectancy challenge and drinking reduction: Experimental evidence for a mediational process. Journal of Consulting and Clinical Psychology, 61(2), 344–353. [DOI] [PubMed] [Google Scholar]
  14. Deluchi M, Costa FS, Friedman R, Goncalves R, & Bizarro L (2017). Attentional bias to unhealthy food in individuals with severe obesity and binge eating. Appetite, 108(1), 471–476. doi: 10.1016/j.appet.2016.11.012 [DOI] [PubMed] [Google Scholar]
  15. DiFeliceantonio AG, Coppin G, Rigoux L, Thanarajah SE, Dagher A, Tittgemeyer M, & Small DM (2018). Supra-additive effects of combining fat and carbohydrate on food reward. Cell Metabolism, 28, 33–44. doi: 10.1016/j.cmet.2018.05.018 [DOI] [PubMed] [Google Scholar]
  16. DiFeliceantonio AG, & Small DM (2018). Dopamine and diet-induced obesity. Nature Neuroscience, 22(1), 1–2. doi: 10.1038/s41593-018-0304-0 [DOI] [PubMed] [Google Scholar]
  17. Falbe J, Thompson HR, Patel A, & Madsen KA (2018). Potentially addictive properties of sugar-sweetened beverages among adolescents. Appetite, 133, 130–137. doi: 10.1016/j.appet.2018.10.032 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Fischer S, Anderson KG, & Smith GT (2004). Coping with distress by eating or drinking: role of trait urgency and expectancies. Psychology of Addictive Behaviors, 18(3), 269–274. doi: 10.1037/0893-164X.18.3.269 [DOI] [PubMed] [Google Scholar]
  19. Gearhardt AN, Corbin WR, & Brownell KD (2009). Preliminary validation of the Yale Food Addiction Scale. Appetite, 52(2), 430–436. doi: 10.1016/j.appet.2008.12.003 [DOI] [PubMed] [Google Scholar]
  20. Goldman MS, Del Boca FK, & Darkes J (1999). Alcohol expectancy theory: The application of cognitive neuroscience In Leonard KE & Blane HT (Eds.), Psychological theories of drinking and alcoholism (pp. 203–246). New York: The Guilford Press. [Google Scholar]
  21. Hales CM, Fryar CD, Carroll MD, Freedman DS, Aoki Y, & Ogden CL (2018). Differences in obesity prevalence by demographic characteristics and urbanization level among adults in the United States, 2013–2016. JAMA, 319(23), 2419–2429. doi: 10.1001/jama.2018.7270 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hall KD, Ayuketah A, Brychta R, Cai H, Cassimatis T, Chen KY, … Zhou M (2019). Ultra-processed diets cause excess calorie intake and weight gain: An inpatient randomized controlled trial of ad libitum food intake. Cell Metabolism. doi: 10.1016/j.cmet.2019.05.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hohlstein LA, Smith GT, & Atlas JG (1998). An application of expectancy theory to eating disorders: Development and validation of measures of eating and dieting expectancies. Psychological Assessment, 10(1), 49–58. [Google Scholar]
  24. Kilbey MM, Downey K, & Breslau N (1998). Predicting the emergence and persistence of alcohol dependence in young adults: The role of expectancy and other risk factors. Experimental & Clinical Psychopharmacology, 6(2), 149–156. doi: 10.1037/1064-1297.6.2.149 [DOI] [PubMed] [Google Scholar]
  25. Lemeshow AR, Rimm EB, Hasin DS, Gearhardt AN, Flint AJ, Field AE, & Genkinger JM (2018). Food and beverage consumption and food addiction among women in the Nurses’ Health Studies. Appetite, 121, 186–197. doi: 10.1016/j.appet.2017.10.038 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Leshner AI (1997). Addiction is a brain disease, and it matters. Science, 278(5335), 45–47. doi: 10.1126/science.278.5335.45 [DOI] [PubMed] [Google Scholar]
  27. Mann LM, Chassin L, & Sher KJ (1987). Alcohol expectancies and the risk for alcoholism Journal of Consulting and Clinical Psychology, 55(3), 411–417. doi: 10.1037/0022-006X.55.3.411 [DOI] [PubMed] [Google Scholar]
  28. Marty L, Chambaron S, Nicklaus S, & Monnery-Patris S (2018). Learned pleasure from eating: An opportunity to promote healthy eating in children? Appetite, 120, 265–274. doi: 10.1016/j.appet.2017.09.006 [DOI] [PubMed] [Google Scholar]
  29. McCarthy DM, & Thompsen DM (2006). Implicit and explicit measures of alcohol and smoking cognitions. Psychology of Addictive Behaviors, 20(4), 436–444. doi: 10.1037/0893-164X.20.4.436 [DOI] [PubMed] [Google Scholar]
  30. McMahon J, Jones BT, & O’Donnell P (1994). Comparing positive and negative alcohol expectancies in male and female social drinkers. Addiction Research, 1(4), 349–365. doi: 10.3109/16066359409005202 [DOI] [Google Scholar]
  31. Meule A, & Gearhardt AN (2014). Five years of the Yale Food Addiction Scale: Taking stock and moving forward. Current Addiction Reports, 1(3), 193–205. doi: 10.1007/s40429-014-0021-z [DOI] [Google Scholar]
  32. Monteiro CA, Cannon G, Moubarac JC, Levy RB, Louzada MLC, & Jaime PC (2018). The UN Decade of Nutrition, the NOVA food classification and the trouble with ultra-processing. Public Health Nutrition, 21(1), 5–17. doi: 10.1017/S1368980017000234 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Montes KS, Witkiewitz K, Pearson MR, & Leventhal AM (2019). Alcohol, tobacco, and marijuana expectancies as predictors of substance use initiation in adolescence: A longitudinal examination. Psychology of Addictive Behaviors, 33(1), 26–34. doi: 10.1037/adb0000422 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Morean ME, Corbin WR, & Treat TA (2012). The Anticipated Effects of Alcohol Scale: Development and psychometric evaluation of a novel assessment tool for measuring alcohol expectancies. Psychological Assessment, 24(4), 1008–1023. doi: 10.1037/a0028982 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Mozaffarian D, Hao T, Rimm EB, Willett WC, & Hu FB (2011). Changes in diet and lifestyle and long-term weight gain in women and men. New England Journal of Medicine, 364, 2392–2404. doi: 10.1056/NEJMoa1014296 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Nijs IM, & Franken IHA (2012). Attentional processing of food cues in overweight and obese individuals. Current Obesity Reports, 1(2), 106–113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Ogden CL, Carroll MD, Kit BK, & Flegal KM (2012). Prevalence of obesity and trends in body mass index among US children and adolescents, 1999–2010. Journal of the American Medical Association, 307(5), 483–490. doi: 10.1001/jama.2012.40 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Pettigrew S (2016). Pleasure: An under-utilised ‘P’ in social marketing for healthy eating. Appetite, 104, 60–69. doi: 10.1016/j.appet.2015.10.004 [DOI] [PubMed] [Google Scholar]
  39. Pursey KM, Collins CE, Stanwell P, & Burrows TL (2015). Foods and dietary profiles associated with ‘food addiction’ in young adults. Addictive Behaviors Reports, 1(2), 41–48 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Racine SE, Hagan KE, & Schell SE (2019). Is all nonhomeostatic eating the same? Examining the latent structure of nonhomeostatic eating processes in women and men. Psychological Assessment. doi: 10.1037/pas0000749 [DOI] [PubMed] [Google Scholar]
  41. Schulte EM, Avena NM, & Gearhardt AN (2015). Which foods may be addictive? The roles of processing, fat content, and glycemic load. PLoS One, 10(2), e0117959. doi: 10.1371/journal.pone.0117959 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Schulte EM, & Gearhardt AN (2017). Development of the Modified Yale Food Addiction Scale Version 2.0. European Eating Disorders Review, 25(4), 302–308. doi: 10.1002/erv.2515 [DOI] [PubMed] [Google Scholar]
  43. Schulte EM, Smeal JK, & Gearhardt AN (2017). Foods are differentially associated with subjective effect report questions of abuse liability. PLoS One, 12(8), e0184220. doi: 10.1371/journal.pone.0184220 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Schulte EM, Smeal JK, Lewis J, & Gearhardt AN (2018). Development of the Highly Processed Food Withdrawal Scale. Appetite, 131, 148–154. doi: 10.1016/j.appet.2018.09.013 [DOI] [PubMed] [Google Scholar]
  45. Schulte EM, Yokum S, Jahn A, & Gearhardt AN (2019). Food cue reactivity in food addiction: A functional magnetic resonance imaging study. Physiology & Behavior, 208, 112574. doi: 10.1016/j.physbeh.2019.112574 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Scott-Sheldon LA, Terry DL, Carey KB, Garey L, & Carey MP (2012). Efficacy of expectancy challenge interventions to reduce college student drinking: A meta-analytic review. Psychol Addict Behav, 26(3), 393–405. doi: 10.1037/a0027565 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Sharkansky EJ, & Finn PR (1998). Effects of outcome expectancies and disinhibition on ad lib alcohol consumption. Journal of Studies on Alcohol, 59(2), 198–206. doi: 10.15288/jsa.1998.59.198 [DOI] [PubMed] [Google Scholar]
  48. Tasevska N, Jiao L, Cross AJ, Kipnis V, Subar AF, Hollenbeck A, … & Potischman N (2012). Sugars in diet and risk of cancer in the NIH‐ AARP Diet and Health Study. International Journal of Cancer, 130(1), 159–169. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Te Morenga L, Mallard S, & Mann J (2013). Dietary sugars and body weight: systematic review and meta-analyses of randomised controlled trials and cohort studies. BMJ, 346, e7492. [DOI] [PubMed] [Google Scholar]
  50. Thanarajah SE, Backes H, DiFeliceantonio AG, Albus K, Cremer AL, Hanssen R, … Tittgemeyer M (2018). Food intake recruits orosensory and post-ingestive dopaminergic circuits to affect eating desire in humans. Cell Metabolism. doi: 10.1016/j.cmet.2018.12.006 [DOI] [PubMed] [Google Scholar]
  51. Thompson FE, Midthune D, Kahle L, & Dodd KW (2017). Development and evaluation of the National Cancer Institute’s dietary screener questionnaire scoring algorithms. The Journal of Nutrition, 147(6), 1226–1233. doi: 10.3945/jn.116.246058 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Walther CAP, Pedersen SL, Gnagy E, Pelham WE, & Molina BSG (2019). Specificity of expectancies prospectively predicting alcohol and marijuana use in adulthood in the Pittsburgh ADHD longitudinal study. Psychology of Addictive Behaviors, 33(2), 117–127. doi: 10.1037/adb0000439 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Wiers RW, Hoogeveen K, Sergeant JA, & Gunning WB (2006). High- and low-dose alcohol related expectancies and the differential associations with drinking in male and female adolescents and young adults. Addiction, 92(7), 871–888. doi: 10.1111/j.1360-0443.1997.tb02956.x [DOI] [PubMed] [Google Scholar]
  54. Wing RR, & Phelan S (2005). Long-term weight loss maintenance. American Journal of Clinical Nutrition, 82, 222S–225S. [DOI] [PubMed] [Google Scholar]
  55. Yang Q, Zhang Z, Gregg EW, Flanders WD, Merritt R, & Hu FB (2014). Added sugar intake and cardiovascular diseases mortality among US adults. JAMA Internal Medicine, 174(4), 516–524. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

tables
NIHMS1060516-supplement-tables.docx (41.3KB, docx)

RESOURCES