Direct and indirect controls of meal size

The brain receives a constant stream of information to guide food intake based on stored energy as well as current energy and nutrient sources (Woods, Reference Woods2005). Thus, the amount eaten within a meal will depend both on the metabolism of previous meals and the currently-available food. However, there is no simple relationship between energy deficit and energy intake. For example, consumers in Western cultures tend to eat their smallest meals after the longest period of energy depletion and their largest meal at the end of the day, in anticipation of this deficit (Weingarten, Reference Weingarten1986). Meal size does not correspond well with the immediately preceding inter-meal interval (Le Magnen, Reference Le1969), suggesting that meal size is highly variable and subject to the influence of many different factors somewhat uncoupled from specific energy deficits (Wiepkema, Reference Wiepkema1971; Lowe & Levine, Reference Lowe and Levine2005). The variability of meal size indicates that there are a number of situational factors that can contribute to the amount of food consumed on any one occasion.

In order to conceptualize the controls of food intake in a clear and objective way, Smith (Reference Smith1996) has proposed that there are direct controls that only relate to the process of food stimuli engaging gastrointestinal mucosal receptors during eating and indirect controls that encompass all other cues. Signals arising from the gut provide both orosensory positive feedback and post-ingestive negative feedback during the eating episode. Orosensory feedback stimulates eating whilst post-ingestive feedback inhibits eating (Smith, Reference Smith2000). During the meal positive orosensory stimulation declines whilst negative feedback increases (Wiepkema, Reference Wiepkema1971). The meal ends when negative feedback exceeds or meets the strength of positive feedback (Smith, Reference Smith1996).

By applying sham-feeding procedures in animal models, in which positive orosensory stimulation is intact but negative feedback is absent, the relative impact of orosensory and gastric feedback in controlling meal size can be determined. The sham-feeding model has also been used to investigate the neuromodulation of direct controls of meal size. For example, it has been shown that central administration of neuropeptide Y, a powerful orexigenic agent, affects meal size by increasing the volume of sucrose solution ingested as well as increasing the frequency and rate of licks in sham feeding compared with normal feeding (Torregrossa et al. Reference Torregrossa, Davis and Smith2006). Thus, neuropeptide Y serves to amplify orosensory stimulation to eat.

Similarly, the application of pyloric cuffs and/or infusions of nutrients directly into the stomach, bypassing orosensory inputs, allows the specific effect of negative gastric feedback to be assessed during a meal. Thus, under these conditions it has been shown that intake of a sweetened milk in rats is reduced by a relatively large volume (12 ml) of either saline (9 g NaCl/l) or milk infusions compared with smaller volumes (3 ml, 6 ml; Eisen et al. Reference Eisen, Davis, Rauhofer and Smith2001). It is interesting to note that in this example reduction of intake does not depend on the nutrient content of the infusion but specifically relates to the larger volume size, suggesting a mechanical threshold for distension cues (Eisen et al. Reference Eisen, Davis, Rauhofer and Smith2001). This finding has been confirmed in human subjects receiving intragastric infusions of milk-based liquid preloads (Rolls & Roe, Reference Rolls and Roe2002). In both lean and obese women increasing volume (from 200 to 400 ml) but not increasing energy (from 834 kJ (200 kcal) to 1668 kJ (400 kcal)) was shown to suppress appetite and food intake. In contrast, intragastric infusions of 425 ml high-fat or high-carbohydrate soup (Cecil et al. Reference Cecil, Francis and Read1999) reduce appetite but not food intake. Correspondingly, in human subjects the distension of either the antrum or fundus, achieved using a gastric balloon, is insufficient even at high volumes of saline (300–800 ml) to reduce food intake, despite increasing subjective fullness and reducing hunger sensations (Oesch et al. Reference Oesch, Ruegg, Fischer, Degen and Beglinger2006). These data indicate that gastric feedback is sufficient to suppress subjective appetite but does not always produce concomitant reductions in food intake. Thus, sham feeding and direct gastric delivery of nutrients allow comparison of the respective roles of positive orosensory stimulation and negative gastric feedback in determining meal size. Evidently, gastric feedback in the absence of orosensory cues is relatively weak at reducing food intake in human subjects (Cecil et al. Reference Cecil, Francis and Read1999; Oesch et al. Reference Oesch, Ruegg, Fischer, Degen and Beglinger2006).

Indirect controls include all other factors that do not directly engage with mucosal receptors, e.g. metabolic, hormonal and cognitive factors and social and other environmental factors. These controls exert their effects on meal size by affecting the potency of direct controls (Smith, Reference Smith1996, Reference Smith2000). To assess the relative potency of direct and indirect influences on food intake, researchers have examined features of the food (sensory characteristics, volume, energy density, portion size), features of the consumer (BMI, dietary restraint, eating habits) and features of the environment (presence of others, presence of competing tasks). It is clear that for human subjects indirect controls can override normal regulatory processes in determining meal size. This situation has obvious theoretical and applied implications for research and practice in understanding appetite regulation. Some of these implications are considered in the present review in the context of cues that promote overconsumption.

Characteristics of the food: the example of variety

During a meal, and long before the food is fully digested and absorbed, satiation processes begin to slow and ultimately to terminate eating. There is therefore a delay between the initiation of eating and the digestion and absorption of nutrients from the gut. Satiation must depend on at least some pre-absorptive mechanisms that increase in strength during the meal and help signal meal termination. It is not clear, however, to what extent satiation is entirely pre-absorptive, since changes in glucose and insulin release occur immediately in response to the sight, smell and thought of food during the cephalic phase of ingestion (Powley, Reference Powley2000). These changes are likely to facilitate satiation. For example, Langhans et al. (Reference Langhans, Grossman and Geary2001) have shown that infusion of glucose or glucose and insulin into the hepatic portal vein during a meal attenuates spontaneous feeding in the rat. Apparently, there are cephalic-phase pre-absorptive processes that provide rapid feedback contributing to satiation. Sensory-specific satiety is a process that begins during eating, but before absorption takes place, and also facilitates meal termination.

Sensory-specific satiety is a phenomenon in which the pleasantness of an eaten food declines during the meal whilst preserving or enhancing the pleasantness of other foods (Rolls, Reference Rolls1986). Thus, the pleasure derived from consumption of a particular food is relatively high in the early stages of the meal and gradually declines during the course of eating. Sensory-specific satiety occurs immediately after eating, in the early pre-absorptive stages of ingestion, and does not rely on gastric feedback. It has been demonstrated when foods are sham fed (Rolls & Rolls, Reference Rolls and Rolls1997; Smeets & Westerterp-Plantenga, Reference Smeets and Westerterp-Plantenga2006), when the food is merely smelled for the same duration as a normal meal (Rolls & Rolls, Reference Rolls and Rolls1997) and for foods with very low energy densities (Rolls et al. Reference Rolls, Hetherington and Burley1988a, Reference Rolls, Hetherington and Burleyb, Reference Rolls, Hetherington and Lasterc). However, it is not known whether sensory specific satiety achieved under sham-feeding conditions will have an impact on energy intake, since not all sham-feeding experiments have offered participants the opportunity to consume a meal after the sham-feeding procedure. Nevertheless, Rolls & Rolls (Reference Rolls and Rolls1997) have reported that in a small sample of ten participants who rinsed (but did not swallow) coffee or orange juice, both sensory-specific satiety and reduced ingestion of the rinsed drink are found. This finding provides preliminary evidence that sensory-specific satiety in the absence of any gastric feedback is sufficient to influence intake, at least in the short term.

Sensory-specific satiety predicts lower intake of the eaten food compared with consumption of other foods. Thus, if offered the same food again this food is less likely to be eaten than if sensorially-different foods are offered (Rolls et al. Reference Rolls, van Duijvenvoorde and Rolls1984). Providing foods that differ only in flavour promotes intake by between 10 and 15%, but presenting foods that differ in flavour, texture and macronutrient content increases energy intake by ⩽40% compared with offering the same food (Rolls & Hetherington, Reference Rolls, Hetherington and Shepherd1989). Thus, greater variety along differing dimensions (sensory and nutritional) promotes greater stimulation of intake. This effect may be related to habituation processes.

In a series of elegant experiments Epstein and his colleagues (for a review, see Raynor & Epstein, Reference Raynor and Epstein2001) have examined the effect of variety on salivary habituation. Salivation in response to the presentation of a food stimulus declines with repeated presentation (Epstein et al. Reference Epstein, Rodefer, Wisniewski and Caggiula1992); however, when a different food is introduced during the session the result is dishabituation to the stimulus (Temple et al. Reference Temple, Kent, Giacomelli, Paluch, Roemmich and Epstein2006). Thus, when children are presented with olfactory cues from a cheeseburger saliva secretion in response to the stimulus declines over seven trials, but this trend is reversed in the group provided with a different food (French fries) in the 8th trial. This change in salivation is mirrored in motivated responding for either the cheeseburger or the French fries (Temple et al. Reference Temple, Kent, Giacomelli, Paluch, Roemmich and Epstein2006). Thus, variety maintains interest in eating by preserving both the salivary response to food and motivated responding to obtain the food, compared with presenting just one food repeatedly.

Variety may stimulate food intake by a number of different mechanisms, including dishabituation of salivary response. Variety maximizes pleasure (positive orosensory feedback) and sustains appetite, whereas consuming a single food facilitates satiation. Introducing many different sensory features within a meal may stimulate eating through extending positive orosensory feedback.

The relative effects of volume (which promotes negative gastric feedback) and variety (which may enhance orosensory positive feedback) on meal size have recently been examined (Norton et al. Reference Norton, Anderson and Hetherington2006) in thirty men and women who on four occasions received a low-volume (females 240 ml, 3·6 kJ/g; males 300 ml, 3·6 kJ/g) or high-volume (females 480 ml, 1·8 kJ/g; males 600 ml, 1·8 kJ/g) tomato soup 30 min before a sandwich lunch with either a single or variety of fillings. Reported hunger and fullness indicate a difference by volume, which suggests that the participants were sensitive to the immediate effects of volume on subjective appetite. However, the satiety quotient (change in subjective ratings of hunger, fullness, desire to eat divided by weight of soup consumed) calculated just before lunch indicates a smaller satiety quotient for the high-volume soup than for the low-volume soup. Thus, the reported satiety quotient for the high-volume soup is weaker, indicating some sensitivity to the energy density of the preloads. Despite the differences in subjective ratings by volume, there is no effect on the intake at lunch, thus intakes following the low-volume and high-volume soup preloads are similar. However, access to a variety of sandwich fillings produces a greater intake (by 14%) compared with providing sandwiches with a single filling regardless of previous volume ingested. Thus, under these conditions the orosensory stimulation provided by variety is more important than the differences in gastric distension produced by alterations in volume.

It has been suggested that the effect of a volume manipulation may also depend on energy density exceeding a certain threshold (Gray et al. Reference Gray, French, Robinson and Yeomans2002). The lower satiety quotient obtained for the high-volume soup in the previous experiment suggests that this threshold was not reached; thus, per g soup consumed, the high-volume low-energy density preload failed to reduce intake at lunch. Under these conditions, increasing sensory variety (positive feedback) has a reliable and marked impact on energy intake in the short term, regardless of previous ingestion of soup. Taken together with evidence from intragastric infusion studies, these data indicate that volume effects will influence food intake only at relatively high volumes (Rolls & Roe, Reference Rolls and Roe2002) and if paired with a relatively-high energy density (Gray et al. Reference Gray, French, Robinson and Yeomans2002).

Variety as a distractor

Introducing sensorially-distinct foods, flavours and textures into a meal may serve to maintain interest in eating. If motivation to eat is enhanced by the presence of varied foods, the meal duration will be extended with a consequent increase in meal size. Sensory specific satiety acts as a brake on intake and is associated with decreased liking for an eaten food, limiting selection and intake of that food whilst directing choice towards foods that differ from the eaten food in sensory characteristics. This process stimulates intake of a variety of different foods. Providing foods that vary along sensory and nutritional dimensions is likely to introduce sufficient novelty to sustain interest in eating and to promote intake, whereas consuming a single food enhances satiation processes, including the decline in the pleasantness of the taste and desire to eat that food.

Consumers who identify the change in pleasantness of the taste of a food as a primary reason for terminating intake tend to eat less than those who rely on gastric fill (Hetherington, Reference Hetherington1996). This finding suggests that if the attention of the consumer is drawn to the change in pleasantness of the eaten food during a meal, this process will facilitate the development of satiation and will limit intake. Attention to what is eaten and memory for recent eating contributes to the regulation of intake (Higgs, Reference Higgs2002, Reference Higgs2005). Multiple attentional targets (whether food- or non-food-related) will delay satiation.

This prediction has been tested in two studies in which participants were asked to consume a snack of either popcorn or chocolate. In the first within-subjects experiment participants attended the laboratory on four occasions for a snack. On one occasion they simply ate their preferred flavour (salty or sweet) of popcorn ad libitum without interruption. In this control condition participants rated the pleasantness of their preferred flavour of popcorn, immediately before and after intake. During the other three experimental sessions participants consumed their preferred flavour of popcorn, rated its pleasantness before and after intake and, in addition, made ratings of pleasantness and desire to eat the eaten food (same food) or a different uneaten food (either congruent or incongruent in flavour to the popcorn) at intervals of 1 min during the snack. Instructions to stop and rate the foods were delivered by audiotape. Interrupting participants to taste and rate a different food during eating increases consumption of their snack by 13% in the congruent condition (where the other food shared similar sensory characteristics) and by 18% in the incongruent condition (where the other food differed in sensory characteristics) compared with intake when attention is specifically drawn to the taste of the eaten food. This same taste condition maximizes focus on the sensory properties of the popcorn, with the congruent and incongruent taste conditions appearing to disrupt this focus. When ratings of pleasantness of the eaten food are compared across conditions, pleasantness declines rapidly in the same condition but remains relatively high in the congruent and incongruent conditions. Thus, increased food intake during congruent and incongruent conditions may be related to the delay in satiation observed in these conditions.

In the second between-subjects experiment participants attended the laboratory on only one occasion. Men and women were allocated to either a food-focus or a food-distraction condition. In the food-focus condition the participants were given chocolate to consume as a snack and were interrupted at 1 min intervals during eating to rate pleasantness and desire to eat that food. In the food-distraction condition the participants were asked to make the same ratings of a different food (cheese crackers) provided alongside the chocolate. Again, by simply introducing a salty food during the consumption of a sweet snack the pleasantness of the snack remains relatively high compared with attending to this food alone during the snack. Introducing variety during an eating episode appears to delay the normal decline in pleasantness associated with sensory-specific satiety (Hetherington et al. Reference Hetherington, Foster, Newman, Anderson and Norton2006b). This finding could explain, in part, why variety increases energy intake.

Variety is likely to stimulate overconsumption through a variety of mechanisms including the process of salivary dishabituation (Epstein et al. Reference Epstein, Rodefer, Wisniewski and Caggiula1992) and extending positive orosensory feedback during the meal (Hetherington et al. Reference Hetherington, Foster, Newman, Anderson and Norton2006b). Short-term experiments suggest only that overconsumption in response to variety occurs within a meal, but it is not clear whether variety is relevant to long-term energy balance. In a study of variety effects in the longer term Stubbs et al. (Reference Stubbs, Johnstone, Mazlan, Mbaiwa and Ferris2001) have studied six lean and six overweight men during a 9 d period with a limited (five), medium (ten) or a high variety (fifteen) of foods offered. In this experiment participants were monitored within a residential environment and the composition of foods was controlled, having identical macronutrient composition and energy density. For all twelve men access to the high variety of foods was found to increase energy intake. Also, the lean men were found to gain weight with the high-variety of foods compared with the medium and limited varieties, and the overweight men tended to lose weight during the study, losing least weight with the high-variety of foods. Thus, providing ad libitum access to foods that differ in sensory attributes enhances energy intake and in some cases produces excess intake and positive energy balance. This result is an important demonstration of how short-term experimental studies can identify a key determinant of energy intake in the laboratory that then translates well to the longer-term context of body-weight regulation. This study also shows very distinctly that not all individuals are equally susceptible to environmental manipulations.

Increasing pleasure during eating

Variety may enhance eating by its effects on habituation and the extension of positive orosensory feedback; however, it is unlikely that variety itself enhances intake. The previously described studies have all involved a variety of highly-palatable foods within a meal or over several days. The implication is that less-liked foods offered over many days or months may not maximize pleasure and therefore intake. In a series of studies conducted over many years by researchers at Natick, MA, USA on army rations (Hirsch et al. Reference Hirsch, Kramer and Meiselman2005) reduced energy intake has been found repeatedly when soldiers receive a variety of ready-to-eat meals in the field compared with the same foods offered in cafeteria settings. A comparison of acceptance ratings between field and control conditions shows higher ratings in the field group but a consistently lower energy intake. However, in the field soldiers discard items they do not like, or trade them for foods they like, which serves to maintain acceptance at a relatively high level despite lowered energy intakes. Reduced energy intake probably reflects long-term monotony and relatively unfavourable situational factors associated with eating in the field. This ‘real world’ example demonstrates the limits of variety in enhancing food intake, the importance of providing frequent changes in menus to promote novelty and the strategies consumers use to maintain the pleasure they derive from food such as trading items.

Another obvious means to extend positive orosensory feedback is the manipulation of palatability. Maintaining the energy and nutrient content of foods whilst manipulating palatability is the only method of determining the relative importance of pleasure over other variables in guiding energy intake and whether changing palatability alone can affect weight regulation. In a series of experiments on rats using an ‘electronic oesophagus’, in which animals can determine the rate of infusion of nutrients directly into the stomach when ingesting a flavoured solution, Sclafani (Reference Sclafani2004) has demonstrated that improving the palatability of the flavoured solution increases both energy intake and weight gain.

In human subjects palatability of individual foods tends to be associated with high levels of energy density, independent of fat levels, and high taste pleasantness is associated with greater levels of energy intake (McCrory et al. Reference McCrory, Saltzman, Rolls and Roberts2006). However, since palatability co-varies with energy density, it is difficult to disentangle the relative impact of each factor on weight gain. However, if energy content is kept the same and foods are made more palatable, as in the rat model, then the relative effects of energy and palatability can be determined.

Yeomans et al. (Reference Yeomans, Lee, Gray and French2001) have investigated this approach by developing a palatable version of a test meal using the simple addition of oregano. This alteration in flavour is sufficient to promote greater food intake. It has been suggested that increasing the palatability of foods may reduce their satiating effects (Yeomans et al. Reference Yeomans, Lee, Gray and French2001; Robinson et al. Reference Robinson, Gray, Yeomans and French2005). Thus, the efficacy of a high-energy fat or carbohydrate preload in reducing subsequent food intake has been shown to be influenced by whether the test meal is bland or palatable. Each of the high-energy preloads reduces intake of the bland meal but not the palatable meal (Yeomans et al. Reference Yeomans, Lee, Gray and French2001). Providing infusions of 300 ml high-energy high-fat soup directly into the gastrointestinal tract fails to inhibit intake of a palatable test meal relative to a bland test meal (Robinson et al. Reference Robinson, Gray, Yeomans and French2005), although these differences are not replicated with the high-carbohydrate soup preload. This outcome suggests that palatability may stimulate eating despite post-ingestive signals associated with satiation and satiety, particularly when these signals are relatively weak, as they are for fat.

Flavours paired with high-energy infusions are less preferred than flavours paired with more-dilute nutrient infusions (Sclafani & Ackroff, Reference Sclafani and Ackroff2004). Thus, there is no simple linear relationship between palatability and increased energy density in stimulating both food intake and weight gain. Nevertheless, in animals flavours paired with nutrients may come to be preferred and this preference can occur even in animals that have been previously fed to satiety on lab chow. Thus, manipulations of palatability may, under certain conditions, disrupt normal satiation processes (Yeomans et al. Reference Yeomans, Lee, Gray and French2001) and override homeostatic mechanisms of energy balance (Sclafani, Reference Sclafani2004). It remains to be seen whether the same can be applied to human subjects over the long term, since the complexity of learning the consequences of eating paired with the diversity of flavours available in the human diet precludes a straightforward comparison. However, the possibility that palatability may impair satiety, and that this process could contribute to overeating and weight gain is worthy of further consideration, particularly given the consumer's efforts to maximize the pleasantness of the eating experience and the palatability of the foods they choose.

Pleasure and alcohol

Alcoholic beverages are often consumed before and with food in order to heighten appetite and to enhance the pleasure derived from the meal. Alcohol has the capacity to influence energy intake and how substrates are utilized (Suter, Reference Suter2005), thus alcohol is an interesting and important component of the diet for body-weight regulation. Energy derived from alcohol is additive to total energy intake, i.e. there is no apparent compensation for alcohol energy within a meal (Poppitt et al. Reference Poppitt, Eckhardt, McGonagle, Murgatroyd and Prentice1996, Reference Poppitt, McCormack and Buffenstein1998). Alcohol also stimulates short-term intake relative to consuming no alcohol (Yeomans et al. Reference Yeomans, Hails and Nesic1999, Reference Yeomans, Caton and Hetherington2003; Hetherington et al. Reference Hetherington, Cameron, Wallis and Pirie2001). Thus, when alcohol is consumed within 1 h of eating there is no adjustment made in intake (additive effect) and it can increase the amount consumed relative to no alcohol (stimulatory effect). It is not clear whether these effects are compensated by reduced intake and/or increased physical activity later in the day or on subsequent days. However, in the short term at least, there is a highly-reliable effect of alcohol on promoting energy intake (Yeomans, Reference Yeomans2004).

Interestingly, when consumers have a variety of foods to select from, alcohol intake favours selection of high-fat palatable savoury foods (Caton et al. Reference Caton, Ball, Ahern and Hetherington2004). It is not known whether alcoholic drinks produce these effects as a function of enhancing pleasure or through learning and expectancy effects. For example, consumers may have long conditioning histories of pairing alcohol with specific types of foods; thus, choice is directed by previous learning or it could be that ingestion of alcohol promotes high-fat high-energy foods as a function of metabolic processes.

To investigate the role of palatability in the ‘aperitif’ effect of alcoholic beverages an experiment has been conducted on twelve lean males (Caton et al. Reference Caton, Bate and Hetherington2006). The objective was to test the effects of alcohol intake on the development of sensory-specific satiety, perceived palatability of foods and food intake. Thus, a comparison was made of the amount eaten from two versions (bland or flavoured) of the same food items (crackers and cheese) at lunch following either a preload with or without alcohol. The no-alcohol preloads consisted of 405 ml chilled lightly-carbonated white grape juice (867·3 kJ), with pure ethyl alcohol (96%, v/v) wiped around the rim to reduce expectancy effects. The alcohol preloads consisted of 375 ml carbonated white grape juice with the addition of 24 g (30 ml) ethyl alcohol (7·5% (v/v); 1508·6 kJ). Pleasantness ratings of the lunch foods were taken alongside a selection of sweet and savoury bland and flavoured foods before and after the drinks and again after the lunch.

It was found that bland foods are rated as less pleasant than flavoured foods; however, alcohol increases intake of both bland and flavoured versions of the lunch compared with intake following no alcohol. Alcohol has no selective effects on perceived palatability and does not affect the development of sensory specific satiety. Thus, alcohol stimulates food intake despite differences in palatability. What are the implications of this robust stimulatory effect of alcohol on energy intake in the long term and on weight regulation?

Most studies of short-term food intake have not followed participants beyond a single meal or have relied on diet records for subsequent meal intake (Mattes, Reference Mattes1996; Westerterp-Plantenga & Verwegen, Reference Westerterp-Plantenga and Verwegen1999). In contrast, Foltin et al. (Reference Foltin, Kelly and Fischman1993) have measured accurately all foods and drinks consumed in a residential setting after dextrose- or ethanol-containing drinks. In this study energy from both beverages was found to be only partially compensated within a 24 h period, although energy compensation on the days after beverage intake was not calculated. It has been estimated that there is a 2 d lag in compensation (de Castro, Reference de Castro1998) and, therefore, any adjustments in intake should be followed for ≥2 d following alcohol. A recent study of lean men (Caton & Hetherington, Reference Caton and Hetherington2005) has compared the intake of a two-course meal following no alcohol, 375 ml glass of red wine before lunch (aperitif) or 125 ml wine served with the first course and 200 ml wine served with the second course (co-ingestion). Energy intake for the remainder of that day and the following 3 d was then measured using diet records. Total energy intake was found to decrease the day after the experiment in all conditions and again on day 4, regardless of alcohol intake. Thus, under these conditions at least, the men adjusted for overconsumption in the laboratory on days 2 and 4. It is possible that for these lean men sufficient adjustment for overeating has a protective effect on weight status. However, these results may differ in older or overweight participants.

There is considerable interest in the relationship between alcohol intake, adiposity and obesity (Wannamethee et al. Reference Wannamethee, Shaper and Whincup2005). Large-scale epidemiological surveys (Arif & Rohrer, Reference Arif and Rohrer2005; Wannamethee et al. Reference Wannamethee, Shaper and Whincup2005) indicate that moderate social drinking is protective against overweight and obesity but that excessive or binge drinking (more than four drinks daily) is associated with a higher BMI (Arif & Rohrer, Reference Arif and Rohrer2005) and with central adiposity (Wannamethee et al. Reference Wannamethee, Shaper and Whincup2005). Similarly, higher alcohol intakes are related to poorer diet quality (Breslow et al. Reference Breslow, Guenther and Smothers2006). However, even this apparently linear relationship masks the observation that the poorest diet quality is associated with the highest intakes of alcohol at the lowest frequencies. Thus, it is not simply the amount of alcohol consumed but the frequency and pattern of drinking as well as the amount consumed that determines its impact on dietary intake and potential to weight gain.

The variable response to alcohol ingestion and body-weight regulation is supported by laboratory investigations (Clevidence et al. Reference Clevidence, Taylor, Campbell and Judd1995) that indicate that weight gain following alcohol is linked to individual differences. The particular case of alcohol highlights the importance of the contribution of this macronutrient to the diet, its effect on stimulating intake and on individual differences in the ability of consumers to adjust energy intake in response to both energy from alcohol and additional food energy in regulating body weight.

Characteristics of the food environment

Human food intake is influenced by social setting; thus, energy intake in the presence of friends or family can increase by between 40 and 70% compared with eating alone (Shide & Rolls, Reference Shide and Rolls1991; de Castro, Reference de Castro1990, Reference de Castro1991, Reference de Castro1994). Eating with friends and family encourages greater food intake both by extending the duration of the meal and by providing a more pleasant context in which to consume meals and snacks (Herman et al. Reference Herman, Roth and Polivy2003). Thus, for example, participants rated a test meal as more palatable when this meal was eaten with a group of others than when eating the same meal alone (Bellisle & Dalix, Reference Bellisle and Dalix2001). Meals eaten alone tend to be somewhat functional, tedious and short. In contrast, eating a meal with others is perceived as an important and enjoyable part of the cultural experience. Indeed, most important social and cultural events tend to revolve around social eating (Rozin, Reference Rozin2005). When consumers eat in groups of familiar others they expect the focus to be on food sharing and the time frame is likely to be much longer than that set aside for eating alone (Herman et al. Reference Herman, Roth and Polivy2003). Similarly, the sorts of foods provided for a social function will differ markedly from those eaten during a solo meal (de Castro, Reference de Castro1994). The social, psychological and environmental factors contributing to social-facilitation effects are highly variable (Hetherington et al. Reference Hetherington, Foster, Newman, Anderson and Norton2006b). However, it could be that eating with others simply stimulates eating as a function of duration, with longer sessions increasing exposure to food cues and providing greater opportunities to eat (de Castro, Reference de Castro1990). It is not possible in most everyday social contexts to identify the relative importance of meal duration or the type of food offered in stimulating consumption.

Thus, social-facilitation effects have been studied in the laboratory so that foods offered can be controlled as well as the duration of the meal. Importantly, key features of the social context can also be manipulated, including the number of diners and their familiarity to the participant. A recent experiment (Hetherington et al. (Reference Hetherington, Anderson, Norton and Newson2006a) has considered the possibility that social eating, like eating in the presence of any distraction, increases food intake partly as a function of drawing the consumer's attention away from the food eaten, thereby extending meal duration and amount consumed. Thus, a comparison was made of intake when eating alone and intake when consumers watched television or had their meal in the presence of familiar or unfamiliar others. Since the focus of interest was in how consumers allocated their attention during the meal (eating, drinking, talking) all meals were videotaped and then coded. A buffet-style meal was offered for lunch to approximate the typical foods eaten during social occasions. The meal consisted of a variety of different foods, including energy-dense palatable foods (bread, cheese, green salad, coleslaw, crisps, cakes). It was predicted that social-facilitation effects would favour selection of the most energy-dense and palatable foods. Thus, the four meal sessions consisted of eating alone with no distractions, eating the same meal with the television switched on or eating with two same-gender friends or two same-gender strangers. Inviting same-gender diners is important since women eat differently in front of men, tending to eat less in order to appear feminine (Mori et al. Reference Mori, Chaiken and Pliner1987).

Under these conditions three key findings emerge: energy intake is higher when eating alone with the television switched on or when eating with friends; meal duration is longer when eating with others (friends or strangers) compared with eating alone; videotapes indicate that attention is largely allocated to the meal when eating alone, but to non-food cues in the other conditions. The finding that when eating alone in the presence of the television (14%) or when eating with friends (18%) the energy intake is higher than when eating alone supports the findings of the previous studies. However, the magnitude of the social-facilitation effect is much less than would be predicted from free-living studies (Bellisle et al. Reference Bellisle, Dalix and de Castro1999; de Castro, Reference de Castro1990, Reference de Castro1991, Reference de Castro1994). This disparity is probably a result of the constraints of the laboratory (limited variety, limited time frame) compared with spontaneous meals in the ‘real world’. Nevertheless, the study shows a selective enhancement of sweet high-fat foods (by >50%) when eating with friends compared with eating alone, indicating that social facilitation favours particular foods (easy-to-eat highly-palatable high-energy snacks) typical of social eating occasions. Such an observation has been reported by Clendenen et al. (Reference Clendenen, Herman and Polivy1994) who have found that eating with friends specifically increases intake of cookies compared with intake when eating with strangers.

It is interesting to speculate on why social contexts may favour overconsumption of high-energy high-fat foods rather than, for example, low-energy-density foods such as salad. It could be ease of eating or the palatability of these foods; it could be that consumers match their eating to what others eat (Herman et al. Reference Herman, Koenig-Norbert, Peterson and Polivy2005) or that when distracted it is more difficult to monitor and limit intake of foods that are typically restricted. Of course, all these factors are likely to play a role in the social-facilitation effect.

Key to social facilitation is familiarity of the other diners (Shide & Rolls, Reference Shide and Rolls1991; de Castro, Reference de Castro1994), since eating with strangers fails to reliably increase intake. Eating with strangers may necessitate greater vigilance than eating with familiar others, since the consumer may self-monitor (impression management) and compare their own intake against what others eat (social norms). When unfamiliar women are paired for lunch to eat pizza, the amount eaten is highly matched between the pair (Herman et al. Reference Herman, Koenig-Norbert, Peterson and Polivy2005), suggesting both high levels of self-monitoring and attention to what the other is eating.

The finding in the experiment (Hetherington et al. Reference Hetherington, Anderson, Norton and Newson2006a) that television viewing increases energy intake by about 14% indicates a similar magnitude of effect to that reported by Bellisle et al. (Reference Bellisle, Dalix and Slama2004). Watching television and eating with friends share the common property of introducing competing pleasurable tasks during eating that distract attention away from processing food cues. Scrutiny of the videotapes of each session indicates that the amount of time spent looking away from the meal is highest when watching television, thus the programme engaged the participant's attention. On the other hand, during social eating much of the time is spent looking away from the meal, with 40% of the time spent talking. Thus, when less attention is paid to the meal the ability to self-monitor is reduced and may interfere with the development of normal satiation. However, the presence of distraction alone is not sufficient to increase food intake, since eating with strangers has the same effect on time spent looking away from the meal, and time spent talking has no impact on energy intake.

Long-term energy balance: implicating situational factors in weight gain

Given the earlier observations, is there any evidence that watching television contributes to weight gain as well as increasing time spent being sedentary? Television viewing is associated with snacking, sedentary behaviour and obesity (Hancox et al. Reference Hancox, Milne and Poulton2004; Viner & Cole, Reference Viner and Cole2005). One study of adolescents (Van den Bulck & Van Mierlo, Reference Van den Bulck and Van Mierlo2004) has estimated that, since many of them eat while watching television, 1 h of television viewing is associated with approximately 653 kJ additional energy intake, and another study (Viner & Cole, Reference Viner and Cole2005) has shown that television viewing at the weekends at 5 years of age predicts obesity at 30 years of age. Thus, television viewing reduces physical activity and provides an opportunity for unmonitored additional energy intake that may track well into later life.

What about eating in a social context; does it contribute to weight gain? One study of the effects of transition from living alone to living with a partner (Kemmer et al. Reference Kemmer, Anderson and Marshall1998; Marshall & Anderson, Reference Marshall and Anderson2002; Anderson et al. Reference Anderson, Marshall and Lea2004) has examined differences in food choice, eating habits, energy intake and body weight. In this study twenty-two couples were followed during the 3 months before and up to 1 year after moving in together. It was found that social eating occasions increase after marriage or cohabitation as couples eat together more often and the number of occasions spent eating with the extended family and friends increases. Couples are more likely to include alcohol with meals and encourage each other to break dietary restrictions (Anderson et al. Reference Anderson, Marshall and Lea2004). There is a shift in the kinds of foods eaten as well as a net increase in total energy intake. Overall, it was found that both men and women gain weight within the first year of moving in together (women 1·5 kg; men 1·7 kg). The short timescale of the study may limit extrapolation to a wider context; however, there is some evidence that marital status influences BMI (Ball & Crawford, Reference Ball and Crawford2005).

A ‘real world’ example of transition that incorporates many different changes in eating habits involves leaving home to go to college. In this transition college students are exposed to a wider variety of foods, they eat together in social groups, regularly drink alcohol and are perhaps more likely to eat while distracted than when eating at home. Several studies (Anderson et al. Reference Anderson, Shapiro and Lundgren2003; Levitsky et al. Reference Levitsky, Halbmaier and Mrdjenovic2004; Lowe et al. Reference Lowe, Annunziato, Tuttman, Didie, Ballace, Riddell, Maille, McKinney and Stice2006) have shown a reliable increase in body weight, known in the USA as the ‘freshman 15’. This phenomenon is widely acknowledged anecdotally but few systematic studies have been conducted. One such study (Anderson et al. Reference Anderson, Shapiro and Lundgren2003) has monitored body weight in 135 first-year men and women over the first 14-week semester and a smaller subsample until the end of the first year. An overall gain in weight of 1·3 (range −3·6–+5·2) kg was found in the first semester, and in the subset of freshmen monitored until the end of the first year the net weight gain was 1·7 (range −6·2–+11·4) kg. These findings are of a similar magnitude to that found in the couples moving in together (Anderson et al. Reference Anderson, Marshall and Lea2004), suggesting some shared experience in the transition process.

Another study Levitsky et al. (Reference Levitsky, Halbmaier and Mrdjenovic2004) that has provided further support for this weight change has reported an average gain of 19 (range −5·9–+8·6) kg in the first semester. Using a self-report questionnaire to investigate predictors of weight gain it was found that the variance in the regression models is largely accounted for by eating at ‘all-you-can-eat’ dining halls, increased snacking and consumption of high-fat snacks. Clearly, situational factors play a profound role in determining how much and what types of food are selected by consumers, and at times of transition (going to college, living as a couple) these changes in eating are sufficient to promote weight gain. It is obvious, however, that not everyone who experiences this sort of transition gains weight. In each of these studies it was found that the majority of participants gain a modest amount of weight; however, some lose weight and others stay the same. The questions that arise are: what predicts susceptibility to these environmental factors promoting overconsumption and weight change; who is most likely to gain weight under these circumstances. To address these questions it is necessary to consider individual differences; however, it is clear that few studies examining situational cues to overeat have identified specific individual characteristics that identify those susceptible to weight gain.

Characteristics of the consumer

If consumers are aware of the tendency for small changes in their dietary intake to produce changes in body weight they may attempt to limit food intake. Restrained eating is the construct used to describe this type of behaviour, i.e. imposing cognitive limits on food intake for the purpose of maintaining or losing weight (for a review, see Herman & Polivy, Reference Herman and Polivy2005). In the case of the weight gain that occurs in the first year of college, it might reasonably be predicted that higher levels of restrained eating would result in less weight gain than in individuals with low levels of restraint. However, Lowe et al. (Reference Lowe, Annunziato, Tuttman, Didie, Ballace, Riddell, Maille, McKinney and Stice2006) have examined weight gain in sixty-nine female first-year students and have related this gain to various measures of dieting and restrained eating. It was found that cognitive restraint scores from validated questionnaires fail to predict weight change; however, women with a history of dieting and with a high level of weight suppression (difference between current and highest-ever weight) are most likely to gain weight during their first year. The average weight gain across the year was found to be 2·1 kg, with this gain being greater (+5 kg) in those currently attempting to diet compared with former dieters (+2·5 kg) and those who had never been on a weight-loss diet (+1·6 kg). Similarly, those students with the highest levels of weight suppression were found to gain more weight over the academic year than those with low weight suppression. Evidently, individuals with a tendency to gain weight, as indicated by high levels of weight suppression, are those most likely to be on a diet and most vulnerable to the situational factors that promote energy intake during their first year. However, it is curious that those claiming to be restricting food intake to lose weight are also those most likely to gain weight. This seeming paradox has been illustrated in a series of studies by Stice et al. (Reference Stice, Fisher and Lowe2004) in which those subjects who score high on restrained eating measures fail to restrict their food intake under a number of different circumstances: in a laboratory-based snack; a laboratory-based healthy meal; a fast-food restaurant; a university dormitory dining hall. In all these different settings it is reasonable to suppose that there would be an inverse relationship between dietary restraint and energy intake; however, this relationship is not observed. Highly-restrained individuals may be restrained for the very reason that they are highly susceptible to food cues and tend to overeat. There are numerous studies (Polivy & Herman, Reference Polivy and Herman1985; Stice et al. Reference Stice, Cameron, Killen, Hayward and Taylor1999) that indicate a strong relationship between dietary restriction and the tendency to binge eat. It is not always clear whether restraint is imposed to counter episodes of overeating or if efforts to restrict food intake increase the likelihood of overeating.

Whatever the direction of causality, there seems to be some circumstances in which restraint is accentuated and the tendency to restrict is enhanced and other circumstances in which restraint is abandoned. For example, Bellisle & Dalix (Reference Bellisle and Dalix2001) have demonstrated that when restrained eaters are asked to focus on the food they are eating a lower intake is produced than when eating while listening to an audiotape. There is a correlation between restraint score and the difference in intake between baseline and distraction. Focusing on food appears to accentuate restraint by producing similar intakes to the baseline condition, but distraction increases intake especially in those individuals with higher restraint scores.

When restrained eaters are faced with cues that remind them of their intention to restrict food intake restraint is heightened, but if attention is narrowed to other cues then this situation promotes disinhibited eating (Mann & Ward, Reference Mann and Ward2004). Dietary restriction is an effortful activity that ‘costs’ the restrained eater (Green & Rogers, Reference Green and Rogers1999). Thus, if restrained eaters are asked to perform a task with a high cognitive load during eating, this additional demand on attention may impair monitoring of food consumption and produces an increase in intake relative to a low-cognitive-load activity (Ward & Mann, Reference Ward and Mann2000).

There is, therefore, no simple relationship between an individual's tendency to restrict food intake and their susceptibility to gain weight in situations that promote overconsumption. It is not clear what specific individual characteristics protect against the obesogenic environment, although it is the case that some individuals are extremely good at resisting even the most systematic attempts at overfeeding. For example, a study of overfeeding (Jebb et al. Reference Jebb, Siervo, Fruhbeck, Goldberg, Murgatroyd and Prentice2006) has been conducted in which six lean healthy men took all meals in a metabolic ‘hotel’ for a period of 17 weeks. Following a 21 d baseline there were three overfeeding phases interspersed by 7 d of free feeding. The men were required to eat +20, +40 and +60% energy requirements, which was achieved by increasing the energy density of the diet through increasing fat content systematically. As expected, the result was a weight gain and, on average, a failure to compensate for overfeeding during the periods of free feeding. However, there was an attempt by at least two men to compensate during the 7 d ad libitum periods, although 7 d may be insufficient for this ability to be fully expressed, and perhaps a longer period of free feeding would have permitted compensation to occur.

Jebb et al. (Reference Jebb, Siervo, Fruhbeck, Goldberg, Murgatroyd and Prentice2006) have demonstrated that individual differences will determine whether weight is gained during systematic overfeeding. Evidently, there are some individuals who are resistant to situational cues to overeat, even when overfeeding is forced. However, it is not obvious what these individual differences are and how they arise. Are susceptible individuals more motivated by food reward or more sensitive to the presence of food cues? Recent imaging studies (Beaver et al. Reference Beaver, Lawrence, van Ditzhuijzen, Davis, Woods and Calder2006) suggest that there is a correlation between sensitivity to reward traits and brain activation in response to palatable food images. There are, therefore, individual differences in the response to situational cues involving food, and these differences may play a role in susceptibility to overeating and weight gain in the long term.