Body Mass Index and stimulus control: Results from a real-world study of eating behaviour

Abstract

Background

Evidence suggests decisions about when, what, and how much to eat can be influenced by external (location, food outlet presence, food availability) and internal (affect) cues. Although the relationship between stimulus control and obesity is debated, it is suggested that individuals with higher BMIs are more driven by cues to eating than individuals in the healthy-weight range (HWR). This study investigates the influence of stimulus control on real-world food intake, and whether stimulus control differs by BMI. It was hypothesised that, compared to those in the HWR, eating among individuals with higher BMIs would be under greater stimulus control.

Method

74 participants (n = 34 BMI < 24.9, n = 40 BMI > 24.9) recorded food intake for 14 days using Ecological Momentary Assessment. Participants also responded to 4–5 randomly-timed assessments per day. Known external and internal eating cues were assessed during both assessment types. Within-person logistic regression analyses were used to predict eating vs. non-eating occasions from stimulus control domains.

Findings

Results support the hypothesis that eating was influenced by stimulus control: food availability, affect, time of day, and location significantly distinguished between eating and non-eating instances (AUC-ROC = 0.56-0.69, all p's < 0.001). The presence of food outlets was significantly better at distinguishing between eating and non-eating instances for those with higher BMIs (compared to individuals in the HWR).

Discussion

Results support the notion of stimulus control in shaping eating decisions. Differences in levels of stimulus control between participants in the HWR compared to those with a high BMI suggest that dietary improvement interventions may be more effective when they are tailored to the individual and consider environmental influences on eating behaviour.

Introduction

Current worldwide figures estimate that in 2016, more than 1.9 billion adults are overweight, of which 650 million are categorised as obese (World Health Organization [WHO], 2018). Overweight and obesity are associated with an increase of a number of chronic diseases such as cardiovascular disease, type 2 diabetes and cancer, responsible for vast costs both to health care systems and society (WHO, 2018; Di Angelantonio et al., 2016; Tremmel, Gerdtham, Nilsson, & Saha, 2017). In order to address what is now considered a growing public health concern (WHO, 2018), it is crucial to understand factors contributing to weight gain.

Although obesity results from a complex combination of biological, behavioural, and environmental factors, one of the most crucial risk factors can be attributed to the health behaviour of the individual (Hunger, Smith, & Tomiyama, 2020; Schoeppe et al., 2016). One such behaviour is dietary intake (Spring, Moller, & Coons, 2012). As such, understanding drivers of dietary intake is crucial for the development of tailored interventions aimed at improving dietary behaviours. The majority of caloric intake occurs for reasons other than restoring energy balance (aka, “hunger”; Brownell & Horgen, 2004). Individuals are prompted to eat by external cues (such as the sight or smell of food through either food advertising or seeing others eat) or internal cues (such as mood) rather than a physiological need to eat (Havermans, 2013). The ability of external cues to influence nondepletion-induced eating is conceptualized as stimulus control (Cornell, Rodin, & Weingarten, 1989; Weingarten, 1985). It has been theorized that stimulus control influences eating behaviour through an automatic processing of (previously conditioned) food-related cues in one's environment (Bilman, van Kleef, & van Trijp, 2017). Such cues are then misinterpreted as signs of energy depletion (i.e., biological hunger), motivating oneself to respond accordingly (Lowe & Butryn, 2007). Stimulus control is especially relevant in terms of understanding the intake of highly palatable, energy-dense foods, as such food items are seen as rewarding and therefore act as a ‘motivational magnet’ driving subsequent behaviour (Berridge, 2004).

Stimulus control is commonly examined in laboratory settings. For example, in a simulated fast-food laboratory, presenting participants with a range fast-food cues (such as smell of French fries cooking and images of high-caloric food items) resulted in a significant increase in caloric consumption compared to caloric consumption in a neutral environment (Joyner, Kim, & Gearhardt, 2017). Simillarly, Prinsen, de Ridder, and de Vet (2013) found that participants were ~3 times more likely to consume a chocolate when the bowl of chocolates was surrounded by discarded chocolate wrappers compared to no wrappers, suggesting that even cues regarding others' eating behaviour can influence one's food choice. Experimental work using cognitive tasks and/or psychophysiological methods (such as fMRI, EEG or eye tracking) has found that individuals with obesity are more attentive to eating-related cues (Hendrikse et al., 2015; Mas, Brindisi, Chabanet, Nicklaus, & Chambaron, 2019), and in turn consume higher amounts of palatable foods compared to those in the healthy-weight range (HWR) (Kakoschke, Kemps, & Tiggemann, 2015; Werthmann et al., 2011). However, it is important to note that numerous laboratory studies have found no difference in responses to food-related cues between individuals in the HWR and those with obesity (e.g., Loeber et al., 2012; Nijs, Franken, & Muris, 2010; Phelan et al., 2011). It has been suggested that the variation in findings may be due to methodological differences (e.g., different measures of attention, hunger, foods presented etc.) (Doolan, Breslin, Hanna, & Gallagher, 2015). As such, the relationship between food cue-reactivity and obesity requires further investigation.

A key concern with laboratory-based studies is their ecological validity; the degree to which they model actual real-world behaviour. A way to overcome this is to triangulate findings from lab studies with those from studies of real-world eating. While observational, Ecological Momentary Assessment (EMA; Shiffman, Stone, & Hufford, 2008) methods are particularly useful for studying eating as they allow individuals to record their food intake in real-time, in participants' natural environments (Ferguson & Shiffman, 2011). Importantly, using EMA methods to examine cue-related eating provides real-world validation of laboratory findings. Such studies have provided support for the relationship between stimulus control and eating. For example, easy availability of foods has been shown to facilitate eating in majority of eating instances in African American women (Zenk et al., 2014). Similarly, the availability of foodstuffs in the environment increased the odds of consuming high-energy snacks and savoury meals after controlling for affect, company and current activity (Schüz, Bower, & Ferguson, 2015). Additionally, seeing others eat increased the odds of consuming a high-energy snack by 39%, and, compared to meals or low energy snacks, negative affect was highest during the decision to consume a high-energy snack in individuals with overweight and obesity (Elliston, Ferguson, Schüz, & Schüz, 2016). Moreover, Schüz, Revell, Hills, Schüz, and Ferguson (2017) found that as Body Mass Index (BMI) increased, the odds of consuming a snack when seeing someone else eat also increased. Such findings suggest that one's BMI may influence how one responds to food-related cues. However, Schüz et al. (2017)'s findings are limited in that only the effects of social cues surrounding eating were examined. Given that cues to eating are likely to be idiosyncratic—that is, that people differ in the type of cues that are most likely to prompt them to eat—it is important to test the effects of a broader range of previously established cues. The present study was designed to examine the relationship between BMI and real-world eating patterns. It was hypothesised that, compared to those in the HWR, eating among individuals with higher BMIs would be under greater stimulus control.