No haste, more taste: An EMA study of the effects of stress, negative and positive emotions on eating behavior

Highlights

• Stress and emotions may lead to increased or decreased food intake.

• Stress and emotions may primarily affect hedonic, but not homeostatic eating events.

• Stress, emotions, taste- and hunger-related eating were assessed in daily life.

• Stress resulted in less taste-related eating, but not hunger-related eating.

• Emotional eating moderated the relationship of negative emotions and taste-eating.

Abstract

Objectives

Stress and emotions alter eating behavior in several ways: While experiencing negative or positive emotions typically leads to increased food intake, stress may result in either over- or undereating. Several participant characteristics, like gender, BMI and restrained, emotional, or external eating styles seem to influence these relationships. Thus far, most research relied on experimental laboratory studies, thereby reducing the complexity of real-life eating episodes. The aim of the present study was to delineate the effects of stress, negative and positive emotions on two key facets of eating behavior, namely taste- and hunger-based eating, in daily life using ecological momentary assessment (EMA). Furthermore, the already mentioned individual differences as well as time pressure during eating, an important but unstudied construct in EMA studies, were examined.

Methods

Fifty-nine participants completed 10 days of signal-contingent sampling and data were analyzed using multilevel modeling.

Results

Results revealed that higher stress led to decreased taste-eating which is in line with physiological stress-models. Time pressure during eating resulted in less taste- and more hunger-eating. In line with previous research, stronger positive emotions went along with increased taste-eating. Emotional eating style moderated the relationship between negative emotions and taste-eating as well as hunger-eating. BMI moderated the relationship between negative as well as positive emotions and hunger-eating.

Conclusions

These findings emphasize the importance of individual differences for understanding eating behavior in daily life. Experienced time pressure may be an important aspect for future EMA eating studies.

Introduction

In prosperous societies, characterized by high availability and public visibility of palatable and energy-rich foods, eating behavior is driven not only by physiological or homeostatic processes like hunger and satiety but also by food hedonics (Berthoud, 2006, Lowe and Butryn, 2007), e.g., to high palatability and cue elicited craving. Especially when it comes to in-between meal snacking or unhealthy eating, hunger does not represent the only and most important reason for consummation (Cleobury & Tapper, 2014). As snacking often involves high fat and sugar products (Cleobury and Tapper, 2014, Ovaskainen et al., 2005), identifying possible aspects promoting this eating behavior seems necessary to optimally tackle the rising prevalence of overweight and obesity (Ng et al., 2014).

Previous literature has emphasized stress and emotions as prominent factors influencing eating behavior. Stress can be defined as a state in which environmental demands exceed an individual’s resources (incl. coping skills), with reactions on cognitive-emotional (i.e., experiencing negative emotions), behavioral, or physiological levels (Dickerson & Kemeny, 2004). Several physiological and psychological accounts have been proposed to explain stress/emotion eating relationships. Regarding physiological accounts, during acute stress, activation of the sympathetic nervous system (SNS, supporting fight-or-flight responses) and the associated noradrenaline release redirect blood-flow away from the digestive system (Torres & Nowson, 2007). Thus, a defensive stress response could be expected to decrease food intake. Effects of hypothalamic-pituitary-adrenal (HPA) activity, however, are more complex since its constituents (corticotropin-releasing hormone, CRH, adrenocorticotropin, ACTH, and cortisol) have combined but also independent effects on food intake and influence a range of other metabolic regulators such as insulin, leptin and neuropeptide Y (Adam and Epel, 2007, Bazhan and Zelena, 2013). Other models consider the availability of highly palatable food as crucial for whether increased or decreased eating occurs in response to HPA axis activation (Bazhan & Zelena, 2013): orexigenic effects are expected in the presence of a high calorie diet, anorexigenic effects in the absence of such a diet. Thus, it is not surprising that empirical results on the stress-eating relationship are markedly mixed: While experimental and questionnaire based studies (Gibson, 2012; Oliver & Wardle, 1999; Stone and Brownell, 1994, Zellner et al., 2006) found individuals to eat less during stressful compared to non-stressful situations, others observed exactly the opposite, namely increased food consumption during stress (Epel, Lapidus, McEwen, & Brownell, 2001; Gibson, 2012; Oliver & Wardle, 1999; Zellner et al., 2006). The latter is supported by a recent review showing that during negative emotions (including stressful states) enhanced food consumption is more likely (Cardi, Leppanen, & Treasure, 2015).

While not all physiological stress-eating mechanisms have been identified, it is likely that such hormonal pathways operate in parallel or in interaction with psychological mechanisms such as learning, coping and emotion regulation. Moreover, large interindividual differences in this domain (see below) and the context dependency of human eating behaviors slow the progression toward conclusive stress-eating models (Adam & Epel, 2007). Hence, present research focuses on interrelating subjective indicators of stress and eating, taking into consideration interindividual differences, stress contexts, time pressure and others. In the next step, such models can then be related to neurohormonal indicators in the quest for integrated biobehavioral models of stress, eating and addiction.

According to the individual difference model (Greeno & Wing, 1994), several between participants’ variables might affect the stress-eating relationship: Eating styles, body mass index (BMI) and gender have frequently been assumed to modify the relationship between stress/negative emotions and eating behavior. In terms of eating styles, restrained eating refers to an effort to restrict food intake in order to control body weight. While highly restrained individuals increase their food intake under stress and negative emotions, unrestrained individuals either decrease their intake (Heatherton, Herman, & Polivy, 1991; Mitchell & Epstein, 1996; Rutledge & Linden, 1998) or show no change (Wardle, Steptoe, Oliver, & Lipsey, 2000). Another eating style, emotional eating, refers to eating in response to negative affect: high scorers on emotional eating scales tend to consume more, while low scorers consume less food following stress induction in the laboratory (Cardi et al., 2015; van Strien, Herman, Anschutz, Engels, & de Weerth, 2012; Wallis & Hetherington, 2009). However, other studies failed to show an influence of emotional eating on food consumption (Adriaanse, de Ridder, & Evers, 2011; Conner, Fitter, & Fletcher, 1999). Lastly, high external eating, referring to eating in response to food cues like its sight or smell, has also been related to increased food intake (Conner et al., 1999; Oliver, Wardle, & Gibson, 2000). Further, with regard to BMI, individuals with obesity tend to consume more when stressed compared to normal weight individuals (O'Connor, Jones, Conner, McMillan, & Ferguson, 2008), although literature is indecisive with respect to a main effect of obesity on food intake (Greeno and Wing, 1994, Torres and Nowson, 2007). Last, gender seems to play a role: women report more perceived stress than men (Cohen, Janicki, & Deverts, 2012) and report higher levels of restrained and emotional eating (Conner, Johnson, & Grogan, 2004). Possibly due to these differences, women seem to be more vulnerable to stress-induced eating (Cleobury and Tapper, 2014, Gibson, 2012, Greeno and Wing, 1994, O'Connor et al., 2008, Zellner et al., 2006).

While informative, the above research has mostly taken place in the laboratory, under standardized and isolated conditions. As a result, we know relatively little about how stress and emotions may impact eating behavior in people’s daily lives. Laboratory studies of eating behavior are also problematic for a number of additional reason: The heightened self-awareness distorts eating under laboratory conditions as revealed by a recent meta-analysis (Robinson, Hardman, Halford, & Jones, 2015), naturalistic social aspects are missing (e.g. dinner with the partner or family), food choice is limited and not matched to individual preferences (Cardi et al., 2015; Edwards, Hartwell, & Brown, 2013; Grenard et al., 2013, Oliver et al., 2000; Wallis & Hetherington, 2009; Wansink, 2004), and measuring only one eating episode does not capture reactivity to varying stress and emotion intensities as they occur in daily life.

To remedy these limitations, the aim of the current study is to examine the effect of stress and negative emotions on eating behavior in daily life using Ecological Momentary Assessment (EMA). EMA conserves naturalistic conditions by recording information as people engage in their usual activities and thereby seems especially important regarding eating behavior (Smyth et al., 2001). Despite these many advantages, the EMA literature on stress, emotions and eating behavior in non-eating disordered populations is surprisingly limited. As one of the few exceptions, O'Connor et al. (2008) reported that employees increase their snacking when experiencing ego-threatening, interpersonal and work-related stress but decrease it under physical stress. Moreover, the stress-eating relationship was more pronounced in individuals with higher levels of emotional, restrained, and external eating styles (as well as disinhibition), as well as in females and individuals with obesity. In other EMA studies, daily hassles were related to more snacking (Zenk et al., 2014), but Evers, Adriaanse, de Ridder, and de Witt Huberts (2013) showed that positive emotions were more predictive of snacking compared to negative emotions.

The present study goes beyond the existing research in several important ways. First, previous studies failed to draw a clear distinction between purely stress- or emotion-related eating alterations. Although stress and emotions are related constructs, their effects on eating behavior may differ profoundly as do the physiological underpinnings of the two. Thus, the present study separately measured negative emotional states and stress. Second, although positive emotional states were found to correlate with increased food consumption (Cardi et al., 2015, Evers et al., 2013), they have hardly been studied in daily life. Hence, the present research included positive emotions to account for this gap. Third, following the above reviewed literature, eating styles, BMI and gender were considered as (between-participant) moderators of the (within-participant) stress/emotions-eating associations. Fourth, an important, but unstudied component in previous studies in naturalistic settings is time pressure: Choosing highly palatable, already prepared foods may be due to insufficient time, energy and planning to purchase and prepare healthier food options during stressful situations (Escoto, Laska, Larson, Neumark-Sztainer, & Hannan, 2012; Jabs & Devine, 2006; Welch, McNaughton, Hunter, Hume, & Crawford, 2009). In order to control for this possible explanation, we measured and analyzed time pressure during eating in the current study. Last and importantly, previously used measures of overt eating behavior (overall calories consumed; macro nutritional composition) confound homeostatic (i.e., more physiological, hunger-driven eating) with hedonic (i.e., taste-driven eating) determinants. Thus, we decided to relate the two key determinants of food consumptions, namely hunger and taste-driven eating to their possible predictors (i.e. stress, negative and positive emotions) rather than predicting actual food consumption, expecting stronger effects of stress and emotions on hedonic taste-based eating. This approach also circumvents the problems with defining snacks (eating for hedonic reasons) and separating them from main meals or meal replacements.

Our methods featured 5 intraday signals across the course of 10 days, allowing for time lagged analyses to support directional interpretations of the effects of stress/emotions on hunger- and taste-driven eating. We hypothesized that taste-driven eating (but not hunger-driven eating) may be enhanced after stressful and highly emotional occasions with this effect being pronounced in female restrained, emotional and external eaters. We had no specific hypothesis regarding BMI, as previous literature has shown inconsistent results as well as regarding time pressure, as this is the first study to our knowledge considering this novel aspect.