Satiation, satiety and their effects on eating behaviour

Authors


Bridget Benelam, Nutrition Scientist, British Nutrition Foundation, High Holborn House, 52–54 High Holborn, London WC1V 6RQ, UK.
E-mail: b.benelam@nutrition.org.uk

Abstract

  • Summary

  • 1Introduction
  • 2Physiological mechanisms of satiation and satiety
    • 2.1 Physiological mechanisms of satiation
      • 2.1.1 Gastric mechanisms of satiation
      • 2.1.2 Intestinal mechanisms of satiation
    • 2.2 Physiological mechanisms of satiety
      • 2.2.1 Gut hormones – episodic signals of satiety
      • 2.2.2 Tonic satiety signals
    • 2.3 The integration of satiety signals in the brain
      • 2.3.1 Anorexigenic pathways in the hypothalamus
      • 2.3.2 Orexigenic pathways in the hypothalamus
      • 2.3.3 Other areas of the brain involved in satiation and satiety
      • 2.3.4 Reward pathways
  • 3Measuring satiation and satiety
    • 3.1 Measuring satiation
    • 3.2 Measuring satiety
      • 3.2.1 Free living vs. laboratory studies
      • 3.2.2 Preload studies
      • 3.2.3 Self-reported measures of satiety
      • 3.2.4 Measuring food intake
      • 3.2.5 Quantifying satiety
    • 3.3 Confounders in satiety research
      • 3.3.1 Physiological confounders
      • 3.3.2 Behavioural confounders
  • 4The effects of foods and drinks on satiety
    • 4.1 Protein and satiety
    • 4.2 Carbohydrates and satiety
    • 4.3 Fibre and satiety
    • 4.4 Intense sweeteners and satiety
    • 4.5 Fat and satiety
    • 4.6 Liquids and satiety
    • 4.7 Alcohol and satiety
    • 4.8 Energy density and satiety
  • 5The effect of external factors on satiation and satiety
    • 5.1 Palatability
    • 5.2 Variety
    • 5.3 Portion size
    • 5.4 Sleep
    • 5.5 Physical activity
    • 5.6 Television viewing and other distractions
    • 5.7 Social situations
  • 6Satiation, satiety and weight control
    • 6.1 Obesity genes and satiety
    • 6.2 Physiological differences in satiation and satiety responses in obese people
    • 6.3 Behavioural differences in the response to satiation and satiety in obesity
  • 7Conclusions

Summary

In the context of the rising prevalence of obesity around the world, it is vital to understand how energy balance and bodyweight are controlled. The ability to balance energy intake and expenditure is critical to survival, and sophisticated physiological mechanisms have developed in order to do this, including the control of appetite. Satiation and satiety are part of the body's appetite control system and are involved in limiting energy intake. Satiation is the process that causes one to stop eating; satiety is the feeling of fullness that persists after eating, suppressing further consumption, and both are important in determining total energy intake.

Satiation and satiety are controlled by a cascade of factors that begin when a food or drink is consumed and continues as it enters the gastrointestinal tract and is digested and absorbed. Signals about the ingestion of energy feed into specific areas of the brain that are involved in the regulation of energy intake, in response to the sensory and cognitive perceptions of the food or drink consumed, and distension of the stomach. These signals are integrated by the brain, and satiation is stimulated. When nutrients reach the intestine and are absorbed, a number of hormonal signals that are again integrated in the brain to induce satiety are released. In addition to these episodic signals, satiety is also affected by fluctuations in hormones, such as leptin and insulin, which indicate the level of fat storage in the body.

Satiation and satiety can be measured directly via food intake or indirectly via ratings of subjective sensations of appetite. The most common study design when measuring satiation or satiety over a short period is using a test preload in which the variables of interest are carefully controlled. This is followed by subjects rating aspects of their appetite sensations, such as fullness or hunger, at intervals and then, after a predetermined time interval, a test meal at which energy intake is measured. Longer-term studies may provide foods or drinks of known composition to be consumed ad libitum and use measures of energy intake and/or appetite ratings as indicators of satiety. The measurement of satiation and satiety is complicated by the fact that many factors besides these internal signals may influence appetite and energy intake, for example, physical factors such as bodyweight, age or gender, or behavioural factors such as diet or the influence of other people present. For this reason, the majority of studies on satiation and satiety take place in a laboratory, where confounders can be controlled as much as possible, and are, therefore, of short duration.

It is possible for any food or drink to affect appetite, and so it is important to determine whether, for a given amount of energy, particular variables have the potential to enhance or reduce satiation or satiety. A great deal of research has been conducted to investigate the effect of different foods, drinks, food components and nutrients on satiety. Overall, the characteristic of a food or drink that appears to have the most impact on satiety is its energy density. That is the amount of energy it contains per unit weight (kJ/g, kcal/g). When energy density is controlled, the macronutrient composition of foods does not appear to have a major impact on satiety. In practice, high-fat foods tend to have a higher energy density than high-protein or high-carbohydrate foods, and foods with the highest water content tend to have the lowest energy density. Some studies have shown that energy from protein is more satiating than energy from carbohydrate or fat. In addition, certain types of fibre have been shown to enhance satiation and satiety. It has been suggested that energy from liquids is less satiating then energy from solids. However, evidence for this is inconsistent, and it may be the mode of consumption (i.e. whether the liquid is perceived to be a food or drink) that influences its effect on satiety. Alcohol appears to stimulate energy intake in the short-term, and consuming energy from alcohol does not appear to lead to a subsequent compensatory reduction in energy intake.

The consumption of food and drink to provide energy is a voluntary behaviour, and, despite the existence of sophisticated physiological mechanisms to match intake to requirements, humans often eat when sated and sometimes refrain from eating when hungry. Thus, there are numerous influences on eating behaviour beyond satiation and satiety. These include: the portion size, appeal, palatability and variety of foods and drinks available; the physiological impact on the body of physical activity and sleep; and other external influences such as television viewing and the effect of social situations.

Because satiation and satiety are key to controlling energy intake, inter-individual differences in the strength of these signals and responsiveness to their effects could affect risk of obesity. Such differences have been observed at a genetic, physiological and behavioural level and may be important to consider in strategies to prevent or treat obesity.

Overall, it is clear that, although the processes of satiation and satiety have the potential to control energy intake, many individuals override the signals generated. Hence, in such people, satiation and satiety alone are not sufficient to prevent weight gain in the current obesogenic environment. Knowledge about foods, ingredients and dietary patterns that can enhance satiation and satiety is potentially useful for controlling bodyweight. However, this must be coupled with an understanding of the myriad of other factors that influence eating behaviour, in order to help people to control their energy intake.

1. Introduction

Maintenance of energy balance and a healthy bodyweight has been critical to human survival, and sophisticated physiological mechanisms exist in the body to maintain homeostasis (the maintenance of a constant internal environment in the body) (Woods & D'Alessio 2008). The control of energy intake is vital to energy balance, and satiation and satiety are part of a complex system of appetite control, which regulates how much we consume. Definitions of satiation and satiety are shown in Box 1.

Box 1 Definitions

Satiation The process that leads to the termination of eating, which may be accompanied by a feeling of satisfaction

Satiety The feeling of fullness that persists after eating, potentially suppressing further energy intake until hunger returns

Over the course of a day, people typically have a number of eating occasions including meals, drinks and snacks. Satiation is important in controlling the amount of energy consumed at each of these eating occasions, while satiety affects the period of time between eating occasions and potentially the amount consumed at the next. Total daily energy intake is a function of both the number of eating occasions that day and their size. Hence, both satiation and satiety affect energy intake.

In the context of the rising prevalence of obesity, it is important to consider the impact of satiation and satiety on energy balance, and whether they can be enhanced in order to facilitate the reduction of energy intake, aiding weight control. However, although this briefing paper will focus on the relationship of satiation and satiety to obesity, it is worth remembering that there are many instances where low energy intake is of concern, for example, in the elderly or those with eating disorders. In these cases, it may be desirable to reduce the effects of satiation and satiety in order to allow greater energy intake.

The factors affecting satiation and satiety from the start of eating to late satiety have been characterised by Blundell et al. (1987) in the satiety cascade, showing sensory, cognitive, post-ingestive and post-absorptive stages shown in Figure 1.

Figure 1.

The satiety cascade showing the influences on satiation and satiety over time (Source: Blundell et al. 1987. reproduced with permission).

As this figure demonstrates, satiation and satiety are initially affected by sensory and cognitive factors including expectations about what is to be consumed, the taste, texture and smell of the food or drink and any associations with previous experience that arise. Once the food or drink reaches the stomach, post-ingestive factors start to take effect. Initially, the distension of the stomach sends signals to the brain, initiating satiation. As digestion continues in the intestines, hormones that promote satiation and satiety are released from the gut. In the post-absorptive stage of the satiety cascade, nutrients themselves are detected by specialist receptors in various sites of the body, including the brain, providing information about nutrient status that also affects satiety (Blundell et al. 1987). In the longer term, satiety may also be affected by signals such as leptin, which convey information about the level of fat storage in the body (Wynne et al. 2005a).

Thus, the body has a complex network of signals involved in the development of satiation and satiety. However, for free-living humans, choices about what and how much to eat are affected not only by internal appetite signals such as satiation and satiety but also by many other factors including the palatability of the food in question, the portion size provided, the time of day and the presence of other people (Bellisle 2003). The amount of energy we consume is completely accounted for by voluntary behaviour, namely the acts of eating and drinking, influenced by physiological, psychological and cultural factors. This is in contrast to energy expenditure, between 20% and 40% of which is under behavioural control via voluntary physical activity (Blundell 2006). Thus, the study of appetite, including satiation and satiety, must take account of both physiological and behavioural evidence in order to gain a full picture of how satiation and satiety affect eating behaviour.

This briefing paper aims to give an overview of how satiation and satiety develop in the body, the factors that affect this and how satiation and satiety interact with external influences to impact on eating behaviour, particularly with regard to excessive energy intake and obesity. The physiological mechanisms of satiation and satiety will first be outlined, followed by a description of the different techniques used to measure satiation and satiety. There is a potential for any food or drink to induce satiety. Thus, it is important to gain evidence as to whether, for a given energy content, particular foods, drinks or their components are more satiating than others. The evidence for the effect of a number of foods, food components and nutrients on satiety is reviewed. Humans may eat when sated and refrain from eating when hungry, so it is clear that internal appetite controls are not the only influence on energy intake. Some of the potential external influences on energy intake are then discussed. Appetite control is one of the potential factors that could affect the risk of obesity (Foresight 2007). In light of the current prevalence of obesity, it may be important to consider whether differences in satiation and satiety or sensitivity to these signals can increase the risk of obesity. There is evidence that this may be the case, and this is outlined in the final section.

This Briefing Paper does not examine functional foods and ingredients and satiety in detail, as these have recently been reviewed elsewhere (Thomas & Chapman 2008). Changes in satiation and satiety in those with eating disorders or other clinical conditions (other than obesity) are also not included. In addition, the area of health claims, satiation and satiety, with reference to the European Commission regulation that came into force in 2007 (EC/1924/2006), is not included in this paper because this is currently the subject of the Institute of Life Sciences International, Europe Appetite Regulation Task Force, which will provide a thorough investigation of the methodologies, relevant food components and appropriate physiological biomarkers to provide scientific substantiation to a satiety claim. This paper does not constitute a systematic review but aims to give a picture of the research in the area of satiation, satiety and eating behaviour.

2. Physiological mechanisms of satiation and satiety

Early animal experiments characterised areas of the brain involved in satiation and satiety by observing the effects of damage in particular locations in the brain, on food intake behaviour. In this way, a number of areas within the hypothalamus were identified as important in the control of hunger and energy balance (Morgane & Jacobs 1969). More recently, the way the body communicates influx, circulation and storage of nutrients and the integration of these signals in the brain to affect satiation and satiety have become more fully understood. An outline of the pathways involved in communication of satiation and satiety between the body and the brain is described in Figure 2.

Figure 2.

A schematic representation of physiological satiation and satiety signalling. When food is consumed, the gastric distension is communicated to the brain via the vagus nerve, which connects the gastrointestinal tract to the brain, initiating satiation. Gut hormones from the stomach and intestine are released when food is consumed and act on areas of the brain involved in appetite. Leptin from adipose tissue and insulin from the pancreas are related to the amount of fat stored in the body and act on the brain to modulate satiation and satiety in the longer term. All signals are integrated in the brain to affect energy intake and expenditure (Adapted from Wynne et al. 2005a).

This section describes the mechanisms by which satiation and satiety signals affect specific areas of the brain to maintain energy homeostasis. It also highlights the possibility that hedonic systems within the brain (i.e. those affected by the pleasurable aspects of food) may interact with homeostatic systems and may override satiation and satiety signals.

2.1 Physiological mechanisms of satiation

Satiation is the feeling of satisfaction that signals eating to stop. The time course of satiation means that factors affecting it must occur early in the satiety cascade, when the food is selected, smelled and eaten and in the first stages of digestion. Satiation appears to be a very basic animal function that even rats with only a hindbrain exhibit (Ritter 2004).

2.1.1 Gastric mechanisms of satiation

When food or drink reaches the stomach, nerves communicate an increase in gastric volume to the brain (Ritter 2004). It appears that gastric distension promotes satiation, independently of nutrient content (Phillips & Powley 2000). In addition, there is evidence that when food is removed from the stomach after ingestion, satiation does not occur. In experiments where cannulas are fitted that allow food to drain from the stomach, animals eat continuously, but quickly reach satiation when the cannula is closed allowing food to fill the stomach normally (Davis & Smith 1990).

There is the possibility that gastric distension could be used as a biomarker of satiation. There are a number of possible indirect methods for measuring gastric distension, for example, measuring changes in water pressure in the stomach, which are outlined in a review of biomarkers in satiation and satiety by De Graaf et al. (2004). Although further work is needed to develop direct markers of gastric distension, this is a possible marker for future research on satiation.

2.1.2 Intestinal mechanisms of satiation

From the stomach, food and drink are released into the small intestine where digestion continues and nutrients are absorbed. It appears that information about the absorption of nutrients can be communicated to the brain and contributes to satiation. Experimental infusions of fat, carbohydrates and proteins directly into the intestine promote satiation. In the case of protein, carbohydrate and fat, digestion to their respective building blocks of amino acids, sugars and fatty acids is necessary in order for satiation to take place (Ritter 2004).

The gut hormone cholecystokinin (CCK) appears to be involved in satiation. The satiating effects of CCK were first demonstrated in 1973, when Gibbs et al. showed that administering CCK reduces subsequent meal size in a dose dependant manner in rats (Gibbs et al. 1973). This has since been confirmed in humans (Kissileff et al. 1981; Muurahainen et al. 1988), although the distension of the stomach by food or drink is necessary for this effect to take place (Lieverse et al. 1995).

CCK is mainly synthesised by the endocrine L cells in the duodenum and jejunum (the beginning and middle of the small intestine) and is rapidly released into the circulation in response to the presence of nutrients in the gut, particularly after fat- or protein-rich meals (Wren & Bloom 2007). CCK's effects on satiety appear to be mediated via receptors in the vagus nerve (the nerve that connects the gut to the brain) and are blocked when this nerve is removed in rats (Smith et al. 1981). The mechanisms by which signals from CCK are integrated within the brain are discussed in section 2.3.

CCK is a potential biomarker for satiation (De Graaf et al. 2004). However, it must be noted that changes in the levels of this gut hormone are only one part of the process leading to satiation and cannot be seen as a direct marker of satiation.

In addition to effects on satiation, CCK also delays gastric emptying and stimulates pancreatic enzyme secretion and gall bladder contraction (Liddle et al. 1985; Moran & Schwartz 1994), thus playing a role in co-ordinating digestion. CCK is also found in the brain, where it acts as a neurotransmitter involved in reward behaviour, memory and anxiety, as well as satiety (Crawley & Corwin 1994). CCK may act synergistically with the hormone leptin, which signals the level of fat storage in the body. This is discussed in section 2.2.2.

In summary, there are a number of variables that influence satiation and thus the amount of food eaten at one sitting. When nutrients reach the small intestine, satiety signalling, which affects the time interval before hunger and the desire to eat returns, is initiated. Although the distinction between satiation and satiety is an important one, they are part of a continuum in the ingestive process, and there may be some overlap between the later stages of satiation signalling and that of early satiety. The next section outlines the mechanisms affecting satiety.

2.2 Physiological mechanisms of satiety

Broadly, satiety is influenced both by short term or ‘episodic’ signals in response to the consumption of food and by longer term or ‘tonic’ signals indicating the levels of energy stores in the body. These act in various ways on the hypothalamus in the brain, which in turn produces signals that affect energy intake and expenditure. This section explores the satiety signals from the body and their effects in the brain.

2.2.1 Gut hormones – episodic signals of satiety

A number of hormones are secreted from the gut to indicate that food has been consumed. These act directly or indirectly on specific areas of the brain to promote satiety. A summary of gut hormones and their actions is shown in Table 1. Ghrelin is the only known gut hormone that causes hunger and, as its suppression is relevant to the onset of satiety, it has also been included for discussion. These hormonal signals are termed as ‘episodic’ signals of satiety as they occur alongside episodes of eating. These are considered separately from tonic signals, which signal the level of energy storage in the body, but it is important to note that there may be interactions between tonic and episodic satiety signalling. The way in which all these signals are integrated by the brain are discussed in section 2.3.

Table 1.  Gut hormones and their actions
NameSite of productionEffect on appetiteMechanismAdditional effects
GhrelinStomach↑ HungerVia ghrelin receptors in the brainLong-term effect on energy balance
Cholecystokinin (CCK)Duodenum and jejunum↑ SatiationVia vagus nerveDelays gastric emptying
Stimulates pancreatic enzyme secretion
Stimulates gall bladder contraction
Acts as a neurotransmitter
Glucagon-like peptide-1 (GLP-1)Intestine and brain↑ SatietyVia GLP-1R in brainIncretin (stimulates insulin production)
Slows gastric emptying
Oxyntomodulin (OXM)Intestine and brain↑ SatietyVia GLP-1R in brainSlows gastric emptying
Via reductions in ghrelin
Peptide YY (3-36) (PYY 3-36)Ileum, colon and rectum↑ SatietyVia Y2 receptors in brainSlows gastric emptying and intestinal transport
Reduces gastric secretions
Pancreatic polypeptide (PP)Pancreas↑ SatietyVia Y5 receptors in brain
Via vagus nerve

Ghrelin is a peptide hormone mainly produced in the stomach and, when administered experimentally to animals and humans, stimulates appetite and increases food intake (Tschöp et al. 2000; Wren et al. 2001). When released, ghrelin acts on receptors in specific areas in the brain. The integration in the brain of signals from gut hormones is described in detail in section 2.3.

Ghrelin levels rise before meals, suggesting that it may play a role in meal initiation in humans (Cummings et al. 2001), although studies have not found that ghrelin levels predict the interval between meals (Callahan et al. 2004). The mechanisms causing release of ghrelin from endocrine cells in the stomach are not yet known. However, the suppression of ghrelin after meals is proportional to the energy intake at the meal (Callahan et al. 2004). In addition, on a per-kilojoule basis, fat appears to be less effective than carbohydrate or protein at suppressing ghrelin (Monteleone et al. 2003; Overduin et al. 2005).

Ghrelin may also play a role in long-term energy balance. In humans, ghrelin levels are inversely correlated with levels of body fatness; that is, they are low in obese subjects (Tschöp et al. 2001), higher in lean subjects (Shiyya et al. 2002) and abnormally high in subjects whose energy intake is chronically restricted, such as those suffering from anorexia nervosa (Tolle et al. 2003).

Glucagon-like peptide-1 (GLP-1) and oxyntomodulin (OXM) are both products of the preproglucagon gene, which is expressed in the brain, pancreas and intestine. In the pancreas, the preproglucagon gene product is processed to produce the hormone glucagon, whereas, in the intestine and brain, GLP-1 and OXM are produced (Murphy & Bloom 2004).

Both GLP-1 and OXM are released into the circulation in response to nutrients in the gut (Le Quellec et al. 1992; Herrmann et al. 1995) and appear to have an effect on satiety. Experimental administration of GLP-1 to humans reduces food intake, decreases ratings of hunger and increases ratings of fullness in normal-weight, diabetic and obese subjects (Flint et al. 1998, 2000a; Näslund et al. 1998, 1999a; Gutzwiller et al. 1999a, 1999b; Toft-Nielsen et al. 1999). There is some evidence that GLP-1 is reduced in obese subjects (Holst et al. 1983; Ranganath et al. 1996; Näslund et al. 1999a) and that levels are restored by weight loss (Verdich et al. 2001). Receptors for GLP-1 (GLP-1R) can be found in areas of the brain involved in appetite, and it is thought that GLP-1 mediates its effects on satiety by acting directly upon these areas (Yamamoto et al. 2003). GLP-1 also slows gastric emptying and modulates gastric acid secretion, contributing to the ‘ileal brake’ mechanism of the upper gastrointestinal tract, a combination of effects that controls the transit of food from the stomach into the intestines, allowing effective digestion (Näslund et al. 1999b).

In addition to its effects on satiety and digestion, GLP-1 is a potent incretin, in that it potentiates the production of insulin (MacDonald et al. 2002). Thus, GLP-1 appears to have multiple roles in promoting satiety, in the ileal brake and in encouraging the release of insulin.

GLP-1 is a potential biomarker for satiety (De Graaf et al. 2004). GLP-1 can be measured from blood samples and is seen to rise for two hours after a meal, compared with the fasting level (Orskov & Holst 1987). More work is needed to establish whether or not GLP-1 is a reliable marker of appetite. A recent study on protein and satiety found that, despite an increase in ratings of satiety after a high-protein vs. low-protein preload, GLP-1 levels were unchanged between the two conditions. Other peptide hormones, peptide YY (PYY) and ghrelin (also discussed in section 2.2.1), were also found to be unaffected (Smeets et al. 2008).

OXM has also been found to reduce food intake in humans (Cohen et al. 2003) and enhance weight loss (Wynne et al. 2005b) and appears to increase energy expenditure (Wynne et al. 2006). OXM can bind to the GLP-1R found in areas of the brain involved in appetite control (Dakin et al. 2001) and reduces plasma ghrelin concentrations in rats (Dakin et al. 2004) and in human subjects. A study using physiological doses, administered intravenously, found that fasting levels of ghrelin were suppressed by 40% compared with those in controls when OXM was administered (Cohen et al. 2003), and this might be a factor in its satiating properties. Similar to GLP-1, OXM also slows gastric emptying (Schjoldager et al. 1989).

Overall, GLP-1 and OXM have similar actions on satiety and gastric emptying, although OXM, as well as the GLP-1R, may work via effects on ghrelin. OXM does not have the incretin effects of GLP-1 but may have more potent effects on weight loss (Wren & Bloom 2007).

The PP fold peptides include peptide YY (PYY), pancreatic polypeptide (PP) and neuropeptide Y (NPY). PYY and PP are produced in the gut and are discussed below. NPY is produced in the brain and is described in section 2.3 on the integration of satiety signals in the brain.

PYY is produced in the L cells of the ileum, colon and rectum (Adrian et al. 1985a; Ekblad & Sundler 2002) and is released into the circulation in proportion to the amount of energy consumed, reaching a plateau after 1–2 hours and remaining elevated for approximately six hours (Adrian et al. 1985a). This release begins before the nutrients reach the distal portion of the gut where PYY is produced, so it appears that PYY secretion may initially be stimulated indirectly, possibly via the vagal nerve (Fu-Cheng et al. 1997). Fasting suppresses the secretion of PYY (Adrian et al. 1985a). The main form of stored and circulating PYY is known as PYY(3-36) (a truncated version of the full peptide) (Grandt et al. 1994).

The administration of PYY has been found to reduce food intake both in rodents (Batterham et al. 2002; Adams et al. 2004; Chelikani et al. 2005) and in humans (Batterham et al. 2003a; Degen et al. 2005). Batterham et al. investigated the effects of PYY(3-36) on both lean and obese subjects and found that, in both cases, energy intake at a buffet lunch two hours after administration was reduced by approximately 30% and that there was also a significant decrease in energy intake during the 24 hours after treatment. In addition, fasting PYY levels were lower in obese than in lean subjects, and body mass index (BMI) correlated negatively with PYY levels. Post-prandial PYY release was also lower in the obese subjects, despite the fact that they consumed more energy (Batterham et al. 2003a). This raises the possibility that a deficiency in circulating PYY could be involved in the development of obesity, although it is currently unclear whether low levels of PYY are a cause or an effect of obesity.

PYY also inhibits gastric emptying, increases transit time through the intestine (Savage et al. 1987) and reduces gastric secretions (Adrian et al. 1985b), indicating it may play a role in the ilieal brake mechanism.

PYY, like other PP fold peptides, binds to Y1-Y5 receptors (Larhammar 1996). PYY(3-36) binds most strongly to the Y2 receptor and it is thought that its effects on satiety are mediated by binding to this receptor in the brain.

PP is mainly produced by the pancreas but also in small amounts in the colon and rectum (Adrian et al. 1976). Like PYY, it is released into the circulation after eating, in proportion to the amount of energy consumed (Track et al. 1980). Experiments have also shown that PP administration reduces food intake in humans. When administered two hours before a buffet meal, PP reduced energy intake by 22% at this meal and throughout the evening and the following morning, leading to a 25% decrease in energy intake over the 24-hour period (Batterham et al. 2003b). PP, like PPY, signals to the brain via the Y family of receptors and is thought to act on Y5 receptors. It may also signal to the brain via the vagus nerve (Wynne et al. 2005a).

Interestingly, both basal and post-prandial levels of PP are suppressed in those with Prader–Willi Syndrome (PWS), a genetic condition associated with hyperphagia (overeating) and obesity (Zipf et al. 1981). Administration of PP reduces food intake in some PWS subjects (Berntson et al. 1993), indicating that an altered PP response may be a component of this syndrome.

2.2.2 Tonic satiety signals

Tonic satiety signals communicate the levels of fat storage in the body to the brain, so that energy intake and expenditure can be balanced to maintain a relatively constant bodyweight. These signals act over the longer term than episodic satiety signals that are activated at each eating occasion.

Leptin is a peptide hormone, mainly produced by the adipose tissue. It is a product of the ob gene, which was first identified and cloned in a severely hyperphagic (over-eating) and obese strain of mutant mouse (Zhang et al. 1994). Circulating leptin levels are proportional to fat mass and BMI, and are reduced by weight loss. However, there is inter-individual variation in the amount of leptin produced at a given percentage of body fat (Maffei et al. 1995).

Leptin appears to influence bodyweight via its effect on energy intake and expenditure. Chronic administration of leptin to rodents causes increased energy expenditure, reduced food intake and loss of bodyweight and fat mass (Halaas et al. 1995). Reduction in leptin levels as a result of weight loss is associated with increased hunger in humans (Keim et al. 1998). Leptin may have a synergistic interaction with CCK, a gut hormone involved in satiation (see section 2.1.2), in that small doses of CCK (which are not effective alone) decrease food intake when leptin is administered at the same time (Barrachina et al. 1997).

A mutation of the ob gene, resulting in a lack of circulating leptin, causes severe obesity in humans (Montague et al. 1997), which can be reversed by administering exogenous leptin both in children (Farooqi et al. 1999) and adults (Licinio et al. 2004). In addition, in those with a heterozygous leptin deficiency (i.e. one rather than two functional copies of the ob gene), there is a greater prevalence of obesity and a higher percentage body fat than in subjects with a fully functional ob gene (Farooqi et al. 2001).

The leptin receptor is expressed in areas of the brain involved in appetite control, and leptin is thought to mediate its effects by acting directly on these receptors (Flier 2004). The effect of leptin in the brain will be discussed in section 2.3. Leptin is transported across the blood–brain barrier via a saturable process (Banks et al. 1996). This appears to be affected by energy balance in that starvation reduces leptin transport across the blood–brain barrier, and refeeding increases it (Kastin & Pan 2000).

Although, as mentioned above, a very small proportion of obesity cases involve impaired leptin secretion, most obese people have relatively high levels of circulating leptin (Maffei et al. 1995), and the administration of exogenous leptin has only a modest effect on bodyweight (Heymsfield et al. 1999; Fogteloo et al. 2003). This indicates that leptin resistance, rather than deficiency, may be associated with obesity. The mechanisms by which this could occur are not yet clear, but impaired transport across the blood–brain barrier may be involved (Kastin & Pan 2000). Blood–brain barrier transport of leptin in mice is reduced by diet-induced obesity (Banks et al. 1999). There is also some evidence in animal models to suggest that neurones in the brain that respond to leptin may become resistant to its effects after chronic exposure to high levels of leptin (Sahu 2002).

Leptin has been suggested as a biomarker for satiety (De Graaf et al. 2004). However, because high leptin levels do not appear to reliably increase satiety, it cannot be assumed that changes in leptin will cause a corresponding change in appetite. Thus, leptin might be useful as a biomarker of satiety in the longer term but may not be useful in subjects with chronically high leptin levels.

Thus, while a lack of leptin causes severe obesity, high circulating levels do not have a similarly dramatic effect in reducing body fat, which has been described as leptin resistance. It is also important to consider that leptin's ineffectiveness in preventing obesity at high levels may be a result of internal satiety signals being ignored in the face of easily available, energy-dense and palatable foods.

Insulin is a metabolic hormone produced by the pancreas. Unlike leptin, which does not rise directly in response to food intake, insulin secretion increases rapidly after meals (Polonsky et al. 1988) and acts to control blood glucose levels. However, over the longer term, levels of plasma insulin are directly related to changes in adiposity, so that levels increase with obesity (Bagade et al. 1967).

In animal models, experimental administration of insulin results in a decrease in food intake and loss of bodyweight (Woods et al. 1979; Ikeda et al. 1986), and inhibition of insulin's actions leads to increased energy intake and weight gain (McGowan et al. 1992). This suggests that insulin contributes to satiety. Obesity and lack of physical activity are associated with insulin resistance, which may be accompanied by dislipidaemia (high plasma triglycerides, and low high density lipoprotein (HDL) (‘good cholesterol’), central fat deposition and high blood pressure (Coppack et al. 2005). It is also possible that obese subjects are less sensitive to the satiating effects of insulin. In a study investigating passive over-consumption of high-fat foods in lean and obese males, hyperinsulinaemia in the obese subjects was associated with a lack of appetite control compared with lean subjects (Speechly & Buffenstein 2000). The results of a study that measured responses to insulin in the specific parts of the brain involved in satiety, suggested that insulin resistance may attenuate the effect of insulin on these areas (Anthony et al. 2006).

Insulin crosses the blood–brain barrier and is thought to act on insulin receptors that are found in the brain (Corp et al. 1986). The signalling pathways activated by insulin are discussed in section 2.3.

Interestingly, it has been suggested that there is cross talk between leptin and insulin. Although leptin levels are associated with fat mass, there appear to be other factors involved, and insulin may play a role in stimulating leptin production. In turn, the leptin receptor is expressed in the pancreatic β cells that produce insulin, raising the possibility that leptin affects insulin production (for a review see Kieffer & Habener 2000).

2.3 The integration of satiety signals in the brain

Both tonic and episodic signals of appetite control act directly through receptors in the brain or indirectly via the nervous system on areas of the brain involved in appetite control. Neurones within these areas express neuropeptides that have downstream effects on energy homeostasis. This section describes how these signals are integrated to affect energy intake and expenditure.

Early animal experiments involving stimulation or damage to different brain regions established the hypothalamus as a centre for appetite control (Morgane & Jacobs 1969), and this picture has since been developed by establishing pathways originating within the arcurate nucleus area of the hypothalamus, which controls feeding and satiety. These pathways can broadly be divided into anorexigenic (inhibit feeding) and orexigenic (stimulate feeding) pathways. Each pathway can be both stimulated and inhibited by signals from the gut, pancreas and adipose tissue. The overall effect is to increase feeding and decrease energy expenditure or vice versa, depending on the availability of nutrients and the levels of energy storage in the body. The pathways involved and how they integrate the tonic and episodic signals that have been described in previous sections are summarised in Figure 3. Other areas of the brain are also involved in satiation and satiety, and these are briefly outlined in section 2.3.3. In addition, reward pathways, conveying the pleasurable qualities of food, may also influence satiety and are discussed in section 2.3.4.

Figure 3.

Circulating hormones influencing energy homeostasis via the arcuate nucleus (adapted from Murphy & Bloom 2004). Continuous lines indicate stimulatory effects, and dashed lines indicate inhibitory effects. AgRP, agouti-related peptide; CART, cocaine-and amphetamine-related transcript; GLP-1, glucagon-like-peptide 1; αMSH, alpha-melanocyte-stimulating hormone; NPY, neuropeptide Y, OXM, oxyntomodulin; POMC, pro-opiomelanocortin; PP, pancreatic polypeptide; PYY, peptide YY.

2.3.1 Anorexigenic pathways in the hypothalamus

Neurones that express the neuropeptides pro-opiomelanocortin (POMC) and cocaine-and-amphetamine-related transcript (CART) have an anorexigenic effect and are stimulated by leptin. POMC is a precursor for the neuropeptide α-melanocyte-stimulating hormone (αMSH), which acts on melanocortin 3 (MC3) and melanocortin 4 (MC4) receptors (Ellacott & Cone 2004). Administration of αMSH to rats inhibits feeding (Rossi et al. 1998) and increases energy expenditure (Pierroz et al. 2002); humans and animals that have a mutant POMC or MC4 gene are hyperphagic and obese (Yang & Harmon 2003). CART is co-expressed with POMC in the hypothalamus. Administration of CART to rats inhibits feeding both under normal conditions and during starvation. Conversely, inhibiting the actions of CART increases feeding (Kristensen et al. 1998).

2.3.2 Orexigenic pathways in the hypothalamus

Neurones that express NPY and agouti-related peptide (AgRP) are orexigenic and are stimulated by ghrelin and inhibited by PYY 3-36, GLP-1, OXM, PP, insulin and leptin. Administration of NPY in animal models causes hyperphagia and obesity (Stanley et al. 1986) and reduces energy expenditure (Billington et al. 1991). NPY is thought to increase food intake and decrease energy expenditure by acting on Y1 and Y5 receptors in the hypothalamus (Gehlert 1999). NPY may also have an inhibitory effect on neurones, producing POMC (Roseberry et al. 2004), therefore having a dual effect of stimulating feeding while inhibiting pathways that reduce feeding. AgRP is an antagonist of the MC3 and MC4 receptors, inhibiting the reductive effects of αMSH on appetite, so injection of AgRP causes an increase in food intake in rats (Rossi et al. 1998).

2.3.3 Other areas of the brain involved in satiation and satiety

Neurones that express POMC/CART and NPY/AgRP connect with other areas in the hypothalamus and convey their signals downstream, producing the effects on energy homeostasis described in section 2.3. A detailed description of the neurotransmitters and pathways involved is beyond the scope of this paper, but see Wynne et al. (2005a) for a review.

The brainstem is also an important area of the brain for energy homeostasis. It receives signals from the gut via the vagus nerve (Sawchenko 1983) regarding gastric volume and information about the presence of nutrients in the gut via signals from CCK (Schwartz et al. 1993; Mathias et al. 1998). The brainstem is thought to affect energy homeostasis both through downstream signals in response to input from the vagus nerve, peripheral circulating signals and via reciprocal connections with the hypothalamus (Wynne et al. 2005a).

2.3.4 Reward pathways

Reward pathways, which are involved in signalling the hedonic qualities of food and drink, may also influence satiety, at least in the short-term. Several signalling systems are involved in the hedonic response to food. Opioids appear to affect food intake. In mice, a lack of opioids removed the reinforcing quality of foods (the amount of work subjects are prepared to put in to obtain food), in the fed but not the fasted condition (Hayward et al. 2002). Similarly, opioid antagonists have been found to decrease palatability but not to affect hunger or satiety in humans (Yeomans et al. 1990). However, in one study, subjects prone to binge eating consumed fewer snacks when given opioid antagonists, despite ratings of hunger and satiety being unaffected by the treatment (Drewnowski et al. 1992). The dopaminergic system is also involved in feeding-reward behaviour. Mice that cannot produce dopamine die because they do not feed, and dopamine replacement restores preferences for palatable foods (Szczypka et al. 2001).

Endocannabinoids appear to act on both the homeostatic and hedonic systems controlling feeding behaviour via the hypothalamus and other areas of the brain. They act on receptors throughout the body, including the brain, adipose tissue, muscle and gastrointestinal (GI) tract, increasing energy intake and fat deposition and decreasing energy expenditure. In the presence of palatable food, they affect appetite by both stimulating the desire to eat and blocking signals to terminate eating (Woods 2007). The ability to block these actions was exploited by using the cannabinoid 1 receptor (CB1) inverse agonist known as Rimonabant (Sanofi-Aventis). This prevented the effects of endocannabinoids, which reduced appetite and aided weight loss (Poirier et al. 2005). However, Rimonabant was withdrawn from the European market in 2008 because of the incidence of psychological side effects, such as depression, associated with its use.

Both homeostatic and hedonic pathways in the brain are complex, and it is not clear whether they operate independently or interact. Indeed, these possibilities are not mutually exclusive and either process may occur under particular circumstances. It is self-evident that the pleasurable sensory aspects of food can override internal satiety signals, and increased sensitivity to hedonic stimuli might be a risk factor for over-consumption and obesity (Blundell & Finlayson 2004).

Key points

  • • Satiation is induced via a number of mechanisms, including gastric distension and the gut hormone CCK.
  • • Satiety is controlled by both episodic (following an eating episode) and tonic (longer-term) signals.
  • • Episodic satiety signals are made by gut hormones. Release of ghrelin, which is associated with hunger, is suppressed after energy intake. A number of gut hormones are released from the gastrointestinal tract when energy is consumed, inducing satiety.
  • • Tonic satiety signals provide information about the amount of fat stored in the body. Leptin is secreted in proportion to the amount of body fat, although there are inter-individual differences in leptin levels at comparable body compositions. Although low leptin levels strongly stimulate energy intake, obese people have chronically high levels of leptin, suggesting that leptin resistance is developed. Baseline levels of insulin also vary according to adiposity, and insulin can impact on satiety.
  • • Specific areas in the brain, particularly the hypothalamus, are involved in integrating signals of satiation and satiety.
  • • In these areas, populations of orexigenic and anorexigenic neurones translate signals of satiation and satiety via downstream pathways to modulate energy intake.
  • • In addition to the pathways controlling energy intake, reward pathways involved in the pleasurable perception of food may interact with homeostatic controls and could also affect eating behaviour.

3. Measuring satiation and satiety

Both satiation and satiety are processes that affect eating behaviour. They can be measured directly via food intake or indirectly via subjective sensations. The methods used and the issues that must be considered when using them are outlined below.

3.1 Measuring satiation

The satiating qualities of foods and drinks can be measured by allowing subjects to consume them ad libitum and monitoring how much is eaten before satiation is reached, compared with a control food. Because the termination of a meal may be affected by factors other than physiological signals, measurement of satiation is usually performed in a laboratory setting, where the environment can be controlled to eliminate confounding effects. These include the variety of foods offered, the dietary restraint (tendency to consciously restrict the amount of food eaten) of subjects or the appeal and palatability of the food (Mattes et al. 2005). Confounders are discussed further in section 3.3.

3.2 Measuring satiety

The measurement of satiety can be achieved through methods that allow subjects to record feelings of satiety or hunger, and/or by measuring food intake directly. Studies on satiety are complicated by the psychological and environmental factors that can conjointly affect eating behaviour, and these may be controlled where possible in studies on satiety. Some of the methods for studying satiety are outlined below.

3.2.1 Free living vs. laboratory studies

In attempting to measure satiety and eating behaviour, there has to be a compromise between being able to measure these precisely and making the results applicable to the ‘real world’. Because of potential confounding by behavioural and environmental factors on satiety and energy intake, studies are often conducted in a laboratory environment. This allows the greatest possible degree of control over the external conditions of the study and this means that the endpoints of interest can be measured accurately.

However, when extrapolating the results of laboratory studies to free-living subjects, where conditions are not subject to the same rigorous control, it may be difficult to determine how relevant these are. In particular, short-term laboratory studies generally make efforts to reduce the effect of learning about the post-ingestive effects of consuming a particular food or drink, for example that hunger returns more quickly after a reduced energy version of a product. These effects might have a meaningful impact on eating behaviour in the longer term (Livingstone et al. 2000).

In practice, free-living studies require subjects to self-report dietary intakes, and these measurements are prone to bias. In particular, the underreporting of energy intakes (Black et al. 1993; Goldberg & Black 1998) and misreporting of macronutrient consumption (e.g. underreporting of fat consumption) (Pomerleau et al. 1999; Goris et al. 2000) can be problematic in obtaining an accurate picture of dietary intakes. In addition, the lack of control over the subjects' environment may make it difficult to interpret results and draw conclusions about the effects of the dietary manipulation in question.

Thus, although it is desirable to conduct studies whose results are relevant to free-living populations, in reality, it is extremely difficult to gain meaningful results in uncontrolled conditions. Therefore, the vast majority of studies on satiety have been conducted in the laboratory under controlled conditions. Many studies use a preload design, which is discussed in the following section.

3.2.2 Preload studies

Studies that aim to measure the effects of a particular variable or variables on the short-term regulation of food intake and appetite often follow a preload design, generally carried out over part or all of a single day. This involves first giving subjects a preload food or drink where the variable of interest is manipulated in order to monitor subsequent effects.

The test and control preloads are matched (as far as possible) for taste, appearance, texture and other sensory qualities that might affect palatability. However, they may be different in energy content, macronutrient or ingredient composition, depending on the hypothesis to be tested. The subjects should be blinded to the differences between control and test preloads if the investigators wish to measure only the physiological effects of the manipulation. If the purpose of the study is to measure both physiological and cognitive effects of the changes made, the subjects may be told how the preloads differ. The formulation of the preload is very important in the ability of the study to measure a difference between the test and control preloads. Energy content appears to be particularly important. Small differences in the energy content of test and control preloads or comparisons of preloads containing relatively small amounts of energy, but differing in their composition, may mean that no effect of the test preload is detected; whereas, if the same variables are changed in preloads with higher energy content, significant effects may be seen (Livingstone et al. 2000)

Typically, visual analogue scales (VAS), a type of self-reported measure of appetite, are used to monitor hunger, fullness and motivation to eat (see section 3.2.3). VAS may be recorded both before and at intervals after the preload is given to monitor changes in reported satiety.

After a pre-determined time interval after the preload, a test meal is given, and energy intake is measured. The time interval between the preload and the test meal is critical to the results of the study because different physiological mechanisms play a role in satiety during the sensory, cognitive, post-ingestive and post-absorptive phases of the satiety cascade (see Fig. 1). If sensory, cognitive or gastrointestinal factors are of interest, then the time delay must be 30 minutes or less. A longer time interval is needed to measure post-absorptive effects on satiety. However, if the interval is too long, differences between the test and control preloads may not be detected (Livingstone et al. 2000). The measurement of food intake in studies of satiety is discussed further in section 3.2.4.

Depending on the study, another meal may be offered to test the effect of the preload over a longer time period and subjects may also be asked to self-report their food intake for an allotted time after the study. Preload studies generally have a crossover design, meaning subjects consume both the control and test preloads on separate occasions, so that effects of inter-individual differences are minimised. Differences in self-reported ratings of satiety and energy intakes after consumption of the control and test preloads are then analysed to establish whether the variable in question had an effect on satiety.

3.2.3 Self-reported measures of satiety

VAS  Originally developed in the field of pain research, VAS are a simple tool to allow subjects to rate their level of hunger, fullness or desire to eat. The scale consists of a line, usually 100 or 150 mm, anchored by an extreme answer to the question posed at either end, for example ‘How hungry are you?’, ‘Not at all hungry’vs.‘As hungry as I have ever felt’ (see Fig. 4). Subjects make a mark on the line, indicating how they feel at that moment, and this is quantified by measuring the distance from the left end of the line to the mark. Because there may be inter-individual differences in the way the VAS are completed, a ‘within subject’ study design, rather than separate groups of control and test subjects, is generally used where subjects act as their own controls.

Figure 4.

Examples of visual analogue scales for hunger/satiety (from Flint et al. 2000).

VAS are generally completed at intervals throughout a study to monitor changes in appetite. Traditionally, VAS have been paper based but can now also be administered by using portable electronic notepads, which have the advantage of logging the results automatically. Electronic notepads can also provide alarms to prompt subjects to complete the VAS (Stratton et al. 1998). However, because the scale appears smaller on the electronic notepad than on paper and subjects appear to be more reluctant to use the extremes of the scale when in electronic form, paper and electronic VAS, although equally reliable, should not be used interchangeably (Stubbs et al. 2001).

VAS are relatively easy to use and process and have been found to be reproducible and valid on a short-term basis, in that the satiety/hunger ratings correlate with energy intake (Flint et al. 2000a). However, there are concerns that this association with food intake is modest and that caution should be taken in interpreting the results of studies by using VAS, especially if this is the only measure of appetite used (Mattes et al. 2005).

There are some issues with the interpretation of VAS. Although subjects make a mark on a continuous scale and this is quantified, it cannot be assumed that a mark of 40 mm along a satiety scale indicates a sensation of half the magnitude of a mark that is 80 mm along the scale. It has also been suggested that subjects may be reluctant to make full use of the scale, either avoiding use of the extremes of the scale or preferring them (Livingstone et al. 2000). Despite these issues with VAS, they remain one of the most widely used tools in research on satiety and are often used alongside measures of food intake (see section 3.2.2 on preload studies).

Category scales  Category scales work on the same principle as VAS – subjects self-report their feelings of hunger or satiety in response to questions. However, instead of a continuous line, numbered categories (usually from 1 to 9) are provided. These may go from the absence of a factor to its extreme (e.g.‘How hungry are you?’ 1 = not at all, 9 = extremely hungry) or represent the extremes of two variables (e.g. 1 = extremely hungry, 9 = extremely full).

As with VAS, there are interpretation issues. Although statements appear on a linear numerical scale, it cannot be assumed that the associated statements have a linear relationship; for example, the difference between statements 3 and 4 might not be equivalent to that between 5 and 6. As with VAS, there may also be a tendency for users to avoid extremes.

Satiety-labelled intensity magnitude (SLIM) scale  The SLIM scale was developed by Cardello et al. (2005) in order to provide better quantitative data from self-reported measures of satiety, using a series of experiments designed to quantify the meaning of a number of statements on hunger and satiety. The responses were used to select and quantify a number of phrases to label a magnitude scale from +100 through zero to −100. When compared with VAS, the SLIM scale was found to be more sensitive and reliable (Cardello et al. 2005). However, this scale requires further development (e.g. translations of terms into different languages) before it can be widely used.

3.2.4 Measuring food intake

Measuring food intake is a common way to assess satiety, but eating patterns may be affected by factors other than internal appetite signals. Humans eat in the absence of hunger for a number of reasons (e.g. the availability of palatable foods, boredom or emotional stress). They may also refrain from eating when hungry if psychological, social or environmental factors prevent them from consuming food (Mattes et al. 2005).

Because eating behaviour can be affected by numerous different variables, studies on satiety tend to be conducted in a laboratory setting where the environment can be closely controlled. This allows subjects' eating patterns to be monitored directly by keeping a record of foods consumed and using a nutritional analysis software to calculate energy and nutrient intakes. The amount of food or drink consumed can also be measured covertly and automatically by using a universal eating monitor (UEM), which consists of a set of weighing scales concealed under a table cloth, connected to a computer that measures the amount of food consumed over time (Kissileff et al. 1980).

In these studies, it is important to consider the kinds of foods offered and the way these are presented to subjects in the laboratory, as this potentially affects how much will be consumed. Obviously, foods must be acceptable to subjects in the study or they will not be consumed, and it is advisable to first conduct pilot studies to test the palatability of test meals. Buffet type meals are often offered to allow subjects to choose from a variety of foods. However, it is known that offering a wide variety of foods can increase energy intake through sensory specific satiety (i.e. a subject may feel sated with by one kind of food but will consume more when alternative foods are offered) (Rolls 1984). This should be taken into account when designing and interpreting studies (Livingstone et al. 2000).

After a study in a laboratory, participants may be asked to record their food intake and/or appetite ratings (using VAS or another self-reported method) when they leave the laboratory, for a period after the study to monitor any subsequent changes in appetite (e.g. for the rest of the day, after a morning spent under laboratory conditions or for the following 24 hours after a day's study). The problems of an uncontrolled environment and misreporting of food intake persist, but the results gained in the laboratory provide a context in which to interpret this. Longer-term studies may provide food to be consumed ad libitum and monitor the amount eaten and may also measure changes in bodyweight. If it is assumed that subjects eat until they are full when eating ad libitum, which may not always be the case, energy intake and any changes in bodyweight can give an idea of the satiating properties of the diet as a whole.

3.2.5 Quantifying satiety

By using the preload design described in section 3.2.2, methods have been developed to quantify satiety.

Satiating efficiency  A test of ‘satiating efficiency’ was developed by Kissileff in 1984 and aims to determine the satiating power of foods according to particular factors such as nutrient composition and energy content. A number of preloads are given, varying one factor at a time to different degrees (e.g. increasing the energy content while keeping the proportions of macronutrients constant). These are followed by a test meal at which energy intake is measured. The energy intake at the test meal is then plotted against the factor in question, and the slope of this line is the ‘satiating efficiency’ per unit of the factor in the food in question (Kissileff 1984).

The satiety index  The satiety index (SI), designed by Holt et al. (1995), involves giving the food to be measured as a preload, then taking VAS every 15 minutes after the preload for the next two hours. Subjects are then given a selection of food and drink to consume ad libitum. A ‘satiety response curve’ is plotted by using the change in satiety ratings on the VAS from baseline over time. This is shown in Figure 5.

Figure 5.

A schematic example of a satiety response curve as used by Holt et al. (1995) to calculate the satiety index (adapted from Holt et al. 1995).

The SI score is calculated by dividing the area under the satiety response curve (AUC) of the test food by the group average AUC of the reference food (white bread), which contains the same amount of energy (1000 kJ), and multiplying the resulting number by 100. Thus, the SI compares foods relative to white bread to allow the satiating properties of foods to be ranked (Holt et al. 1995).

The satiety quotient  The satiety quotient (SQ), developed by Green et al. (1997), provides a measure of a food's ability to decrease motivation to eat. It is calculated by measuring the motivation to eat immediately before a test preload then at intervals afterwards. The SQ is calculated from the difference between these pre- and post-consumption motivations divided by the weight or energy content of the food or drink being tested. This gives a measure of the rate at which the motivation to eat returns, by weight of food or by energy content (Green et al. 1997).

These methods provide a way to compare the effects of different foods on satiety. However, they are only as reliable as the measures taken to calculate them and, as discussed in section 3.2.3, self-reported measures to indicate satiety have potential problems, which need to be considered when using satiating efficiency, SI and SQ. In addition, comparisons using these methods can only be made either within a study or between studies that have exactly the same study design. A comparison between studies with different designs would be misleading, as the values obtained are dependent on this.

3.3 Confounders in satiety research

Research into satiety generally aims to investigate the effects of a particular food/nutrient on eating behaviour (e.g. whether fibre-rich foods enhance satiety, causing a decrease in energy intake). Indeed, energy intake can be used as a measure of satiety. However, this is complicated by the fact that other factors can affect how much people eat and how satisfied they feel after eating. If researchers do not wish these to impact on results, then they need to be controlled for. Some possible confounders in satiety research are outlined below, subdivided into physiological and behavioural factors. Some studies may measure the effects of these factors on their results, and, in these cases, the factors of interest will not be controlled for.

3.3.1 Physiological confounders

Bodyweight  Obese and lean subjects may respond differently to tests of appetite and energy intake, not least because their energy requirements are different – the energy requirements of obese subjects are generally higher than those of lean subjects. There is also some evidence to suggest that there may be differences in some aspects of physiological appetite control between lean and obese people that could potentially affect the results of studies on satiety (see section 6). For these reasons, studies may choose to include lean, overweight or obese subjects only. In some cases, however, the differences between these populations may be of interest, and satiety and energy intake may be compared in lean, obese or overweight subjects.

Age  The age of subjects may impact on satiety responsiveness, in particular to sensory specific satiety. This is the phenomenon whereby satiety is reached for a particular food or type of food, but, when different foods are presented, appetite returns, stimulating further energy intake (Rolls 1984). This is discussed further in section 5.2. Sensitivity to sensory specific satiety appears to decline with age (Rolls & McDermott 1991), and this may be particularly important if the test meal offered is in the form of a buffet where subjects are offered a variety of foods.

Gender  The gender of participants in studies may affect their appetite and energy intake. First, simply because women have lower energy requirements than men, they are likely to eat less, and this must be considered if energy intake is an endpoint of the study.

In addition, if female participants of childbearing age are included, it is important to take the menstrual cycle into account. Women's energy intake fluctuates within the menstrual cycle, with two distinct periods of elevated energy intake that appear to be related to hormonal fluctuations. The difference in energy intake between these two phases and baseline intakes (364 kJ or 87 kcal/day) has the potential to confound research on satiety that uses energy intake as an indicator, and this should therefore be noted and controlled for (Lissner et al. 1988).

3.3.2 Behavioural confounders

Habitual diet, alcohol and physical activity  The habitual diet, alcohol consumption and physical activity of subjects can also affect satiety and eating behaviour. Attempts may be made to ensure that participants are not significantly different with regard to these factors before studies commence, and subjects may be excluded on this basis or asked to refrain from physical activity and alcohol consumption the day before a study. Control of diet before a study may be particularly important in those who may not be in energy balance before the study, for example, obese subjects or those on a weight loss diet (Livingstone et al. 2000). However, Gregersen et al. (2008) found that standardising the diet of subjects prior to the study did not improve the reproducibility of ad libitum energy intakes and actually led to an increased energy intake on the following day. This issue may require further clarification.

Dietary restraint  Dietary restraint refers to the tendency to restrict food intake in order to maintain or lose bodyweight and is often associated with dieting behaviour. Dietary restraint may be associated with ‘disinhibition’, where control over eating is lost, resulting in binge eating. Obviously, this type of behaviour in the subjects of a study on satiety could affect the results, and investigators commonly use questionnaires such as the Dutch eating behaviour questionnaire (Van Strien et al. 1986) or the three-factor eating questionnaire (Stunkard & Messick 1985) to assess potential dietary restraint behaviour in study subjects. These use a series of questions on aspects of eating behaviour, providing a score indicating the degree of dietary restraint exhibited by the subject. Those who score above a threshold level are commonly excluded from studies on satiety that wish to exclude the effect of dietary restraint on results.

Prior knowledge and beliefs about test foods  Most studies on satiety manipulate the content of test meals covertly in order to control the influence of beliefs or knowledge about the foods or drinks in question on subsequent intake. However, this is very difficult to do comprehensively, and subjects may still be able to detect differences between the control and test meals. If the study requires a number of visits to the laboratory to test different formulations, subjects will become accustomed to the conditions of the experiment and what is expected of them, and this may also affect their response (Livingstone et al. 2000).

Studies that have tested the effect of prior knowledge about differences in the energy or fat content of foods have found that these can affect both subsequent energy intake and feelings of satiety. When subjects are given identical preloads but told that one is high in fat and the other low in fat, more energy is consumed after the preload labelled as low fat (Caputo & Mattes 1993; Shide & Rolls 1995). In addition, in a study of restrained eaters, sensations of hunger and fullness were affected by the perceived energy content of a liquid preload even when this information was incorrect (Ogden & Wardle 1990). Thus, although difficult to do, it is essential that these factors are controlled as far as possible, unless the effect of prior knowledge and beliefs about the foods in question is one of the variables to be measured.

Key points

  • • The effects of foods and drinks on satiation can be measured by the amount consumed ad libitum.
  • • Satiety can be measured directly via recording energy intake and indirectly by using self-reported measures of appetite.
  • • Visual analogue scales (VAS) are the most commonly used method of self-reporting appetite.
  • • Many studies on satiety use a preload design where a food or drink of interest or a control is given, generally followed by VAS at selected intervals. Then, after a predetermined period of time, a meal is given, and energy intake is measured.
  • • A number of physiological and behavioural factors can potentially confound studies on satiety and, as far as possible, are controlled for in studies that wish to exclude their effects.
  • • In order to create a controlled environment, satiety studies are generally conducted in a laboratory. However, this can make it difficult to extrapolate results to the ‘real world’, where conditions are not controlled for in this way.

The effect of other people  It has been observed that people consume more energy when eating with others, although the nature of the relationship with these other people may modulate this effect (De Castro 1994, see section 5.7). For this reason, subjects in satiety studies generally eat alone or in separate booths to minimise the social effects of the presence of others on eating behaviour.

4. The effects of foods and drinks on satiety

It is possible for anything we eat or drink to affect satiety, so it is important to understand whether, relative to their energy content, different foods or components of foods may have consistently differing and meaningful effects on satiety and subsequent energy intake, at a feasible level of consumption.

This section looks at the effects on satiety of particular nutrients and food components, for example, fat and fibre, and other variables such as energy density and the degree of hydration (liquid vs. solid). Satiety, rather than satiation, is the main focus, as this is where the most research has been conducted. It should be noted that this section does not systematically review the evidence in this area but aims to present an overview of the effects of different foods and drinks on satiety.

Studying the effects of one variable in food or drink while keeping others constant is inherently difficult, especially if researchers do not want the differences to be discernable to subjects. For example, adding fibre to a food can also decrease its energy density and palatability, both of which could affect satiety. Conversely, adding fat could have the opposite effect. Laboratory studies control test foods, so that only the factors of interest are varied. However, this may generate differences in the attributed effects of specific macronutrients between laboratory situations (where many confounding factors are largely controlled) and the real-food environment (where they are not).

Although studies generally make efforts to control for confounding variables, it may not always be possible to keep all factors constant, and this should be considered when looking at the results. Many short-term studies in this area follow the preload design, monitoring appetite ratings and energy intake after a test food or drink. In the longer term, changes in bodyweight in subjects eating ad libitum may be measured to draw inferences about the effects of particular dietary patterns on satiety. Often these studies do not include measurements of appetite, so, although it is assumed that subjects who are not consciously restricting their intake will eat to fullness, caution must be exercised in drawing conclusions about satiety in these cases. Finally, epidemiological studies may find associations between particular dietary patterns and bodyweight, and, although these cannot demonstrate a causal effect of a particular diet on satiety, they can lend weight to a line of evidence.

4.1 Protein and satiety

Many studies have investigated the effects of protein on satiety, and most but not all have found that, at sufficiently high levels, protein has a stronger effect on satiety than equivalent quantities of energy from carbohydrate or fat. Most studies investigating the effects of protein on satiety have followed a preload design, where the protein content of the preload is varied and the effects on subsequent self-reported ratings of appetite and/or energy intake are measured.

A review of studies on protein, satiety and weight loss by Halton and Hu (2004) looked at both short-term and longer-term studies on satiety, energy intake and bodyweight change. The studies were selected on the basis of making a comparison between higher and lower protein preloads or diets monitoring satiety, energy intake and bodyweight change. The authors highlighted that the methodologies used were too different for a systematic comparison to be made and that the review was qualitative in nature. Out of 14 short-term studies, 11 found that the higher protein preload significantly increased ratings of satiety, and 8 out of 15 studies found that the subsequent energy intake was significantly lower in the higher protein condition than in the control (Halton & Hu 2004). The test period in the different studies ranged from 1 to 24 hours, and the proportions of macronutrients in the test and control preloads also varied with 29% to 100% protein in the high protein test and a variety of higher carbohydrate or fat preloads for the control. The form of the preload also varied, from mixed meals to single drinks (Halton and Hu 2004). Another study, (not included in the Halton and Hu review) found that, in the short-term (over 5 hours), when the macronutrient content of a test meal was varied but the energy density was kept constant, there were no differences in appetite ratings or subsequent energy intake (Raben et al. 2003).

Halton and Hu (2004) also looked at the effect of high-protein diets on weight loss. Of the 15 studies identified, seven found a significant weight loss after a high-protein diet compared with the control. Of the studies selected, all 5 of those that allowed ad libitum intake found a significant weight loss, whereas only 2 of the 10 studies that provided isocaloric high- vs. low-protein diets had significant weight loss results (Halton & Hu 2004). Although appetite ratings were not taken during these studies, it may be that increased satiety on the high protein diets vs. the control diets mediated lower energy intakes in the ad libitum studies, resulting in weight loss.

A subsequent study that looked at the effects of both isocaloric and ad libitum high-protein diets lends support to this theory. Subjects were given a diet that provided either 15% or 30% energy from protein (carbohydrate was kept constant at 50% of energy and fat varied from 35% to 20% of energy). For the first four weeks of the study, the high- and low-protein diets were isocaloric, and subjects reported significantly higher ratings of satiety on the high-protein diet than on the low-protein diet. For the following 12 weeks, the macronutrient proportions of the diet remained constant, but subjects were allowed to eat ad libitum from foods provided. This resulted in a spontaneous reduction in energy intake on the high-protein compared with the low-protein diet of 1852 kJ (441 kcal) per day on average and an average weight loss at the end of the study period of 4.9 kg. This reduction in energy intake did not appear to cause a reduction in satiety according to self-reported appetite ratings (Weigle et al. 2005).

Lejeune et al. (2006) conducted a study on protein and satiety in respiration chambers, allowing energy expenditure to be assessed. For four days, subjects were fed either an adequate-protein diet (10% energy from protein) or a high-protein diet (30% energy from protein), which were isocaloric. VAS measurements showed that satiety was significantly increased and hunger was significantly reduced on the high-protein diet, compared with the adequate protein diet, despite energy intakes being the same. Sleeping metabolic rate and diet-induced thermogenesis (DIT) were significantly higher on the high-protein than on the adequate-protein diet (Lejeune et al. 2006). These results confirm those of a similar study performed by this research group in 1999 (Westerterp-Plantenga et al. 1999). In a subsequent review on the effect of protein intake on weight management, Westerterp-Plantenga et al. (2007) highlight the relationship between DIT and satiety, both of which are increased on high-protein diets. It is suggested that the associated increase in body temperature and oxygen consumption when on the high-protein diet may enhance satiety (Westerterp-Plantenga et al. 2007).

A feature of some popular high-protein diets that are also very low in carbohydrate is the induction of ketosis. This is the state where ketone bodies are generated from fat stores in response to reduced glucose availability when carbohydrate intake is very low. An intervention study with obese male subjects by Johnstone et al. (2008) was designed to test whether high-protein, low-carbohydrate (LC), ketogenic diets were more satiating and caused greater weight loss than high-protein, medium-carbohydrate (MC), non-ketogenic diets. The study compared a high-protein, LC, ketogenic diet (30% energy from protein, 4% energy from carbohydrate) with a high-protein, MC, non-ketogenic diet (30% energy from protein, 35% energy from carbohydrate). The energy density of the LC and MC diets were the same, but the carbohydrate and fat contents varied reciprocally. The study had a crossover design, so that subjects consumed both diets. The foods for each diet were provided, and the subjects instructed to eat ad libitum for four weeks. Hunger and energy intakes were found to be significantly lower on the LC ketogenic diet than on the MC non-ketogenic diet, and subjects lost significantly more weight on the LC ketogenic diet. Other measures of appetite were tested, such as fullness and prospective consumption (how much subjects thought they could eat), but were not found to be significantly different between the two diets (Johnstone et al. 2008). It is not possible to conclude whether it was the ketosis on the LC diet or the difference in fat content between the two diets that caused the reduction in hunger and energy intake, and the relevance of ketosis to satiety may warrant further investigation. It should be noted, however, that there are safety concerns about very-high-protein diets and that caution should be exercised in promoting them (Eisenstein et al. 2002).

Some studies have investigated the short-term effects of different protein sources on satiety. Uhe et al. (1992) measured the relative satiating effects of protein in beef, chicken and fish over a period of three hours. VAS measures of satiety were found to be significantly higher after subjects consumed fish than beef or chicken (Uhe et al. 1992). Subsequent energy intake was not measured. Similarly, Borzoei et al. (2006) also looked at the satiating effects of beef and fish and found a non-significant increase in satiety and a significant decrease in energy intake at a subsequent meal after the fish compared with the beef (Borzoei et al. 2006). However, both studies were relatively small (6 and 23 subjects, respectively) and included only lean men. The mechanism by which fish protein might exert greater effects on satiety is not known, but the authors hypothesised that it might be a result of differences in amino acid content or in the slower rate of digestion of fish protein.

Mycoprotein, which is a high-protein food produced from a fungal source, has also been tested for its effects on satiety. Burley et al. (1993) and Turnbull et al. (1993) compared the effects on satiety of a mycoprotein-based meal with those of a chicken meal, with the same protein content. In both studies, subsequent energy intake was lower after the mycoprotein than after the chicken meal. Although the meals were matched for energy and protein, it should be noted that the mycoprotein meal was higher in fibre, which may have affected the satiety response. It is therefore not possible to draw conclusions about the specific the effects of protein on satiety in this case (Burley et al. 1993; Turnbull et al. 1993). Williamson et al. (2006) compared the effects on satiety of a mycoprotein preload with those of chicken and tofu, with matched protein contents. Energy intake at the test lunch was reduced after both the tofu and mycoprotein preloads, compared with the chicken preload. There was no significant difference in this reduction in energy intake between the tofu and mycoprotein preloads. There were also no significant differences between any of the three preloads on the ratings of appetite or energy intake at the evening meal. However, this does mean that subjects did not compensate for eating less at lunch time (after the tofu or mycoprotein preloads) by eating more at the evening meal (Williamson et al. 2006).

Different types of isolated protein added to meals have also been investigated for their potential effects on satiety. Lang et al. (1998) looked at the effects of egg albumen, casein, gelatin, soy protein, pea protein and wheat gluten on satiety, using a preload design, and found no significant differences in their effects on appetite (Lang et al. 1998). The doses used in this study were extremely high (70 g of protein per meal), and this was followed up with a similar study using casein, gelatin and soy protein at lower doses (50 g and 25 g). However, again, there were no significant differences found between the different types of protein (Lang et al. 1999).

Hall et al. (2003) compared the effects of drinks containing very large amounts (48 g per serving) of whey vs. casein (both milk proteins) and found that energy intake was reduced by 19% at a subsequent meal by the whey compared with the casein preload (Hall et al. 2003). Anderson et al. (2004) compared the effects of 45–50 g of protein in liquid preloads from whey, soy protein, egg albumen, sucrose and a control (water) on subsequent energy intake. Whey and soy protein suppressed energy intake at a meal provided one hour later, but egg albumin and sucrose did not, resulting in a greater energy intake overall (preload plus meal). Whey was more effective at suppressing subsequent energy intake than soy, resulting in compensation for the energy in the preload of 96% and 59% respectively, after the whey protein preload and the soy preload.

In a separate experiment, the effect of 50 g of soy protein was compared with a combination of 25 g of soy protein and 25 g of carbohydrate (either glucose or amylose) on energy intake at a meal one hour later. Soy protein alone caused a significant reduction in energy intake, but when the protein content was reduced and carbohydrate was added, there was no significant reduction in energy intake (Anderson et al. 2004). Overall, evidence suggests that the source of protein itself, at levels feasible in foods, does not have a large and consistent effect on subsequent appetite and food intake.

In summary, there is relatively consistent evidence that energy from protein, in a sufficient dose, has a greater effect on satiety than an equivalent amount of energy from carbohydrate or fat, in the short-term, although it is important to note that this effect has not always been observed in studies that have matched the energy densities of macronutrient preloads. This appears to translate into larger amounts of weight loss in longer-term studies when ad libitum high-protein diets are compared with lower-protein diets. This may be a result of a higher satiating effect of protein, although appetite measures are often not taken in these studies, making it difficult to confirm this.

Differences in study design make it difficult to pinpoint the optimum dose or percentage of energy needed to observe significant effects of protein on satiety. Anderson and Moore (2004) suggest that at least 50 g of protein in a food or meal is necessary to see a significant effect on satiety, but that there is not currently sufficient information to describe a dose–response relationship. There is currently no formal definition of ‘high protein’ as a percentage of energy in a meal or diet. In a review of the safety and efficacy of high protein diets, Eisenstein et al. (2002) suggested that protein intakes higher than 25% energy should be defined as ‘high’ and over 35% energy as ‘extremely high’ based on the US dietary recommended intakes which give 10–35% as the acceptable range of protein intake (Eisenstein et al. 2002).

4.2 Carbohydrates and satiety

Carbohydrates are a diverse group that includes mono-, di- and oligosaccharides and starches made from long chains of glucose. The rate at which sugars (mono and disaccharides) and starches are digested, absorbed and metabolised are different because the longer chain length of starches increases the time needed for digestion compared with sugars. In addition, there are a number of monosaccharides (e.g. glucose, fructose, galactose) that are combined to make disaccharides, such as sucrose, lactose and maltose, and there are also variations in the metabolism of different monosaccharides. Therefore, the effect of a carbohydrate on satiety depends on the form of the carbohydrate and other aspects of the food (e.g. fibre content) from which it is delivered.

In the 1950s, the glucostatic theory of appetite regulation was developed by Mayer (1953), which hypothesised that blood glucose levels determined appetite, initiating energy intake when low and causing satiety when increased. Although the effect of nutrients on appetite is now understood to be far more complex (Stubbs 1996), glucose concentration does play a role in appetite control. Because the brain depends heavily on glucose as a source of energy, glucose levels are tightly controlled, and there are multiple sites in the body where glucose levels are detected. This information is integrated in the brain, feeding into a number of neuronal pathways, including those affecting energy intake (Marty et al. 2007). Thus, appetite regulatory systems appear to be directly affected by glucose in addition to the hormonal signals discussed in section 2.

A number of sugars have been investigated for their effects on satiety. As with studies on protein, a preload design is commonly used, and, in order to assess the effects of individual sugars, the preload is often a drink with the sugar in question added. With regard to sucrose, (a disaccharide of glucose and fructose), review studies have found that 25 g as a single dose appears to be the lower detection limit in terms of inducing satiety (Anderson & Woodend 2003), and 50 g or more of sucrose has been shown to increase satiety and reduce food intake in the following 20–60 minutes in adults and children, although the reduction in energy intake does not fully compensate for the additional energy from the sucrose preload (Anderson 1995). It should be noted that the studies included in these two reviews were relatively small and short-term and so conclusions about the longer-term effects of sucrose consumption on satiety or energy intake cannot be drawn.

In a review of the effects of refined carbohydrates on satiety and mood, Hammersley et al. (2007) emphasised the difficulty in disentangling the physiological effects of refined carbohydrates on satiety and the psychological effects of sweetness. Results from the short-term studies on carbohydrate and satiety were inconsistent, and the authors of the review highlighted the fact that a number of studies had not blinded subjects to the content of the test foods/drinks. With the longer-term studies on carbohydrate and satiety included, it was also difficult to draw overall conclusions because of the influence of the inherent palatability of sweetness on food intake (Hammersley et al. 2007).

Reid et al. (2007) investigated the effects of drinks containing sucrose vs. those containing aspartame on satiety and energy intake for a period of four weeks in 133 free-living female subjects with healthy bodyweights. Subjects were provided with drinks each week but were blinded as to their composition. The drinks provided were actually diet and sugar-sweetened versions of a well-known brand of soft drink. However, subjects were told that the drinks had been created especially for the study to minimise any associations subjects might already have with this particular brand. The study had a crossover design so that subjects were tested with both the sucrose and aspartame containing drinks and acted as their own controls. The weekly allocation (comprised of four 250 ml bottles to be consumed over the course of the week) of sucrose-sweetened drinks provided 1800 kJ, whereas the aspartame-sweetened drinks provided 170 kJ/week. No effect was seen on ratings of satiety, but the timing of appetite ratings was not designed to detect short-term changes in appetite after the drinks. In the sucrose condition, subjects reduced their energy intake, partly compensating for the additional energy from sucrose. However, because compensation was approximately 50%, consumption of sucrose-containing drinks resulted in an increased energy intake compared with the baseline measurement. Consumption of the aspartame-containing drinks resulted in a small reduction in energy intake, compared with the baseline (Reid et al. 2007).

Glucose and fructose preloads have also been found to reduce subsequent energy intake (Anderson & Woodend 2003). There is some uncertainty regarding the relative effects of glucose and fructose on satiety. Studies comparing the two tend to be small and short-term and have yielded conflicting results as to which is the more satiating (Anderson 1995; Anderson & Woodend 2003). Glucose and fructose are absorbed and metabolised differently and may act differently on satiety pathways. Glucose itself is detected in the brain and causes insulin release, which also acts on appetite centres in the brain to induce satiety (see section 2.1.2). The way in which fructose feeds into satiety pathways is less clear, and some studies have found that leptin levels are reduced and ghrelin levels are not suppressed in response to fructose ingestion (Teff et al. 2004). This might be expected to lead to impaired satiety although, because there are many other factors involved in satiety-signalling pathways, this cannot be assumed.

A number of studies have compared the effects of sucrose and high-fructose corn syrup (HFCS) on satiety. HFCS is typically comprised of 55% fructose and 45% glucose, although other proportions exist, and so is not that dissimilar to sucrose in terms of its composition. During the storage of soft drinks, sucrose (50% glucose and 50% fructose) is 90% inverted to glucose and fructose within three months (Hein et al. 2005). So, in this case, the composition of sugars in drinks sweetened with sucrose compared with HFCS may not be that different. A number of studies have investigated whether drinks sweetened with HFCS compared with sucrose have different effects on satiety, and a significant difference between the two types of sweetener has not been found (Akhavan & Anderson 2007; Monsivais et al. 2007; Soenen & Westerterp-Plantenga 2007). For more information on liquids and satiety, see section 4.6.

Aside from sugars, much research on carbohydrates and satiety has focused on the impact of glycaemic index (GI) or glycaemic load (GL). GI is a measure of the capacity of carbohydrate-containing foods to raise blood glucose compared with a standard, usually glucose or white bread. GI is defined as the incremental area under the blood glucose curve of a test food containing 50 g of carbohydrate, expressed as a percentage of the response to a portion of the reference food containing 50 g of carbohydrate, taken by the same subject on a different day (FAO/WHO 1998). A high-GI food is one that is absorbed from the gastrointestinal tract relatively rapidly, resulting in a sharp peak in blood glucose, rather than in the more gradual rise over time seen with low GI foods.

GL is calculated by multiplying the available carbohydrate (the total carbohydrate minus fibre) in a serving of the product by its GI and then dividing by 100. This was introduced for providing a measure of glycaemic response that was relevant to a realistic portion of a food, rather than a constant amount of carbohydrate (Jenkins et al. 2002). For example, a carrot has a high GI but a low GL, as the amount of carbohydrate per portion is relatively small. When considering the effects of GI or GL, it is important to note that there are a number of different factors that influence GI and GL, including fibre content and fat or protein content, all of which may have independent effects on satiety. In particular, low GI foods may also be high in fibre, which can enhance satiety (see section 4.3), and this can be a confounding factor. Thus, the literature on GI/GL and satiety is complicated by the fact that, unless the composition of test foods is very carefully controlled, any observed effects on appetite or energy intake may not be a result of differences in glycaemic response per se, but due to other factors, such as macronutrient composition, fibre content and palatability.

A review by Bornet et al. (2007) examined a number of studies that had investigated the relationship between the glycaemic response to foods and its impact on satiety and weight management. Confounding factors were taken into account when selecting studies for this review. However, the authors commented that there was difficulty in distinguishing the effects of GI from those of fibre. Both short-term studies (one day or less) with liquid carbohydrates (e.g. glucose or fructose in water) or foods and longer-term studies (2 weeks – 6 months) were included. Twenty-six short-term studies using foods were selected. While some studies were inconclusive, 16 out of the 26 studies concluded that satiety was increased with low-GI vs. high-GI foods. The authors of the review concluded that, despite some possible confounding factors, such as the fibre content and palatability of the foods, there was evidence that low-GI foods had greater satiating effects than high-GI foods in the short-term. However, seven longer-term studies were identified, and their results did not suggest that the increased satiety seen in the short-term studies had an effect on energy intake or bodyweight over the longer time periods studied (Bornet et al. 2007).

A study by Alfenas & Mattes (2005) compared ad libitum high- and low-GI diets over a period of 8 days. A selection of high-GI and low-GI foods were provided, and subjects ate only those foods for the duration of the study. The high- and low-GI foods were matched for macronutrient composition and palatability, but the low-GI diet was higher in fibre. Appetite ratings were taken on days 1 and 8 after breakfast and lunch, and energy intake was recorded throughout the study. No differences were observed in appetite ratings or in energy intakes between the high-GI and low-GI diets (Alfenas & Mattes 2005). This suggests that, under carefully controlled conditions, the GI of foods does not affect satiety or energy intake.

Das et al. (2007) conducted a randomised controlled trial (RCT) that looked at the effects of two energy-restricted diets, one high- and one low-GL. All foods were provided for the first six months of the trial, and subjects were guided by individual eating plans prepared during discussions with dietitians. The researchers designed the diets to avoid confounding factors, such as the palatability and variety of the diet. Satisfaction, with the quantity and types of food provided, hunger, and desire to eat foods that were not included in the study were measured during the first three months of the study. There was a significant decrease in satisfaction with the foods provided, within the high GL group, but not within the low-GL group, but there were no significant differences over time between the high- and low-GL groups. Both groups decreased their energy intakes, compared with baseline, and lost weight and body fat, but, again, there were no significant differences between the high- and low-GL groups (Das et al. 2007).

Aston et al. (2008) performed an RCT measuring the effect of a higher- vs. lower-GI ad libitum diet on appetite, dietary intake and bodyweight using a crossover design (subjects were tested with both diets and acted as their own controls). The diets were matched for energy content, macronutrient composition and fibre content. No differences were found between the higher and lower GI conditions for any of the endpoints measured (Aston et al. 2008). Overall, the evidence suggests that, when potential confounding factors are controlled, the GI or GL of foods does not have a significant impact on satiety.

Thus, carbohydrates are a diverse group of compounds, and it is difficult to make generalisations about their effects on satiety. Although carbohydrates have been suggested to be intermediate in their impact on satiety – less satiating than protein, but more so than fat – comparisons between the satiating power of carbohydrates and fats have provided inconsistent results (see section 4.5).

4.3 Fibre and satiety

The term ‘dietary fibre’ encompasses a variety of compounds that reach the colon undigested, including insoluble fibres such as wheat bran, soluble fibres from oats and fruits, resistant starches and oligosaccharides. These may be found naturally in foods or isolated and used as functional ingredients (for more information, see Lunn & Buttriss 2007; Buttriss & Stokes 2008). The effect that a type of fibre has on satiety depends on its physical properties when eaten and its physiological effects in the gut.

The literature on dietary fibre and satiety was reviewed by Slavin and Green in 2007. They looked at studies using high-fibre diets, whole foods and isolated fibres that can be added to foods. With regard to high-fibre diets, two studies in this review looked at the effect of a high-fibre breakfast on subsequent energy intake at lunch (Burley et al. 1987; Silberbauer et al. 1996). Neither study detected an impact of fibre on energy intake but Burley et al. found that the high-fibre breakfast caused greater fullness in subjects' appetite ratings. Another study compared the effects of ad libitum high- and low-energy density diets on satiety, energy intake and eating time over five days. The high-energy density diet was achieved by using products that were high in fat and sugar and by avoiding fibre. The low-energy density diet was made up of high-fibre foods and was low in fat. Subjects consumed half the amount of energy per day on the low- compared with the high-energy density diet, and eating time was 33% longer on the low-energy density diet (Duncan et al. 1983). However, it is not possible to discern from this study whether the results were specifically due to the fibre content of the low-energy density diet or to the effect of energy density overall or to some other factor.

Studies on a number of whole foods, such as fruit, vegetables, oats, barley and bread, were included in the review. Most, but not all, studies showed an effect of fibre on satiety and subsequent energy intake. The amount of fibre was important, and larger doses were more effective at promoting satiety and reducing energy intake. Other factors also appeared to play a role, such as the extent to which the food was refined or processed, the particle size in grain-based foods and energy density (Slavin & Green 2007).

Many types of isolated fibre were included in this review, and some, but not all, had a significant effect on satiety and/or energy intake. Generally, those studies that used high doses of fibre (e.g. more than 10 g in one dose or 30 g over the course of a day) had more positive results. However, in some cases (e.g. when using polydextrose), even high doses had no effect. The more viscous fibres, such as pectin, psyllium and guar gum, were the most effective at increasing satiety and had an impact on satiety even at relatively low doses. Slavin and Green (2007) suggested that viscous fibres could prolong the digestion of food and absorption of nutrients, extending the time available to stimulate pre- and post-absorptive mechanisms of satiety. The bulking effect of fibre can increase chewing time and gastric distension, promoting satiation. Adding fibre also reduces energy density, which may also promote satiety (see section 4.8) (Slavin & Green 2007).

Novel fibres, which form gels in the stomach because of a chemical reaction with stomach acid, have been developed and tested with respect to their effect on satiety and energy intake. Hoad et al. (2004) tested the effects of drinks containing alginate fibres (one that formed a gel in the stomach and one that did not) compared with drinks containing guar gum (a fibre whose viscosity is unaffected by stomach acid) and a control (a sweetened milk drink) in 12 overweight and obese subjects. All the drinks containing added fibres were significantly more satiating over a 240-minute period than the control, and there was a trend for increasing satiety with increased viscosity (the gel-forming alginate being the most viscous) (Hoad et al. 2004).

Pelkman et al. (2007) compared the effects on energy intake of drinks providing two different levels (1 and 2.8 g) of a gelling pectin-alginate fibre in a drink, compared with a control drink without added fibre, in 29 overweight and obese subjects. The drinks were consumed twice a day over 7 days, and energy intake at an evening meal was recorded. The 2.8 g dose of pectin-alginate caused a decrease of approximately 10% in energy intake at the evening meal, which was not compensated for by an increase in energy intake later. The 1 g dose reduced energy intake in some, but not all subjects (Pelkman et al. 2007). Paxman et al. (2008) tested a gelling alginate fibre vs. a control in 68 men and women for 7 days. Daily consumption of the fibre resulted in a 7% reduction in energy intake in these subjects (Paxman et al. 2008). Thus, this novel action of gelling fibres appears to be effective in inducing satiety and reducing energy intakes, but more studies may be needed to confirm these effects. It should also be noted that some consumers might find products containing these fibres unpalatable (Hoad et al. 2004).

4.4 Intense sweeteners and satiety

Intense sweeteners provide a sweet taste in foods or drinks without delivering energy. There has been some debate surrounding their effects on satiety and energy intake, in particular whether they help to reduce energy intake overall or disrupt appetite regulatory systems, resulting in a loss of appetite control and overeating. Because there are a number of different intense sweeteners that can be used in a variety of products, this is a complex area and effects may depend on the sweetener used and the product containing it.

Drewnowski and Bellisle (2007) reviewed this area and highlighted that, as energy density is an important determinant of energy intake (see section 4.8), the effect of intense sweeteners on energy density of foods is an important factor. In those soft drinks in which the energy content is provided by sugars, the replacement of sugar with intense sweeteners produces a large reduction in energy density, to almost zero. In solid foods where energy is also provided by protein and fat, replacement of sugars with intense sweeteners has less impact on energy density. The studies investigating the impact of intense sweeteners on satiety and energy intake, reviewed by Drewnowski and Bellisle, had mixed results, with some finding increases in appetite and/or energy intake, some a decrease, but the majority finding no significant effect on these parameters. Differences in study design make it difficult to reach any overall conclusions about the effect of intense sweeteners on satiety, but it appears that intense sweeteners do not enhance satiety. The reduction in energy intake because of the reduced energy density of foods/drinks containing intense sweeteners does not seem to be compensated for by an increase in subsequent energy intake. However, this could only be helpful in reducing energy intake within the context of a controlled diet (Drewnowski & Bellisle 2007).

Appleton and Blundell (2007) investigated the effects of intense sweeteners and sugars on habitual high (HC) and low consumers (LC) of drinks containing intense sweeteners. In LC subjects, subsequent energy intake was increased in response to the sweet tastes of both preloads containing intense sweeteners and sugar, whereas, in HC subjects this increase was not observed. This suggested that HC subjects had learnt to dissociate sweetness from energy provision, whereas LC subjects had not (Appleton & Blundell 2007). This study indicates that previous experience with intense sweeteners may affect appetite and energy intake after their consumption.

4.5 Fat and satiety

Dietary fat affects satiety by slowing gastric emptying, stimulating the release of satiating gut hormones and suppressing the release of ghrelin (Little et al. 2007) (see section 2.2.1). However, it has been suggested that the effect of fat on satiety is weaker than that of either protein or carbohydrate (Rolls et al. 1988; Westerterp 2004). This view is not supported by all studies, and it is important to note that an increase in the fat content of a food or diet tends to increase palatability and energy density, both of which can also affect satiety and energy intake (Rolls & Hammer 1995). Some of the studies that have investigated the effects of fat on satiety are outlined below.

Blundell et al. (1993) investigated the effects on subsequent energy intake of the addition of either a carbohydrate or fat supplement of 1.52 MJ (362 kcal) to a standard breakfast in lean male subjects, 90 and 270 minutes after eating the breakfast. The carbohydrate supplement resulted in a reduction in energy intake after 90 minutes but not after 270 minutes, but the fat supplement produced no reduction in subsequent energy intake. Although the fat and carbohydrate supplements were matched for energy content, the higher energy content per gram of fat than of carbohydrate results in a higher energy density (kJ/g; kcal/g) in the fat supplement. So it is not possible to discern whether the results in this study were due to the specific effects of fat on satiety or differences in energy density.

In a further study, obese female subjects were fed either a high- (4.12 MJ/985 kcal) or low-(2.21 MJ/527 kcal) energy lunch, and then given an evening meal where they were allowed to choose from either a range of high-fat foods (minimum 50% energy from fat) or a range of high-carbohydrate foods (minimum 50% energy from carbohydrate). There was a non-significant increase in energy intake at the evening meal after the low- vs. the high-energy lunch of 0.59 MJ (139 kcal). However, exposure to the choice of high-carbohydrate or high-fat foods caused a significant difference in energy intake with subjects consuming an average of 2.84 MJ (677 kcal) and 5.59 MJ (1336 kcal) on the high-carbohydrate and high-fat meals, respectively (Blundell et al. 1993). Thus, providing foods with a higher fat content stimulated passive over-consumption of energy at the evening meal. However, again, increasing the fat content of foods also increased their energy density compared with the high carbohydrate foods, and it might be energy density, not the fat content per se, that led to the increased energy intakes observed in this study.

Stubbs et al. (1996) looked at the effects of breakfasts high in protein (HP), carbohydrate (HC) or fat (HF) on appetite ratings and energy intake at a test lunch, five hours later and for the rest of the day (until 11 pm). The breakfasts were matched for energy density and palatability. They found that hunger was highest after the HF breakfast compared with the HP and HC breakfasts, and that the HP breakfast was the most effective at suppressing hunger throughout the day. However, these differences did not translate into significant differences in energy intake either at the lunch or during the rest of the study day (Stubbs et al. 1995b).

Bell and Rolls (2001) compared the effects of meals containing varying amounts of fat, matched at two different levels of energy density. This created six different experimental conditions: lower-energy density with a low-, medium- or high-fat content or higher-energy density with a low-, medium- or high-fat content. Thirty-six subjects attended the laboratory on six occasions (to test each experimental condition) and ate breakfast, lunch, dinner and an evening snack. Ratings of appetite and daily energy intakes were measured. No significant differences were seen in appetite ratings, but energy intakes were significantly higher (by 20%) on the higher-energy density than on the lower-energy density diets. When energy density was matched, the fat content of the diets did not affect energy intake, indicating that it was the energy density and not the fat content that influenced satiety (Bell & Rolls 2001).

Saltzman et al. (1997) conducted a study with seven twin pairs, investigating the effects on energy intake of a high-fat vs. a low-fat diet eaten ad libitum over 11 days. The two diets were matched for energy density, and no significant differences were detected in energy intakes between the two diets (Saltzman et al. 1997)

When high-fat diets are compared with lower-fat diets in the longer term, it has been found that subjects consume less energy and lose more weight on low fat diets, without any restriction on the amount of food eaten (Glueck et al. 1982; Lissner et al. 1987; Kendall et al. 1991). Glueck et al. took a number of daily appetite ratings and found no significant differences between the high-fat and low-fat conditions. Lissner et al. and Kendall et al. did not measure appetite. These studies suggest that humans do not detect a reduction in energy when the fat content of the diet is reduced and thus do not compensate by eating more. However, although palatability of the high-fat and low-fat diets was matched in these studies, energy density would have been lower in the low-fat condition, so it is not possible to conclude whether it was the reduction in the fat content of the diet specifically or the lower energy density of the low-fat diets that led to the reduction in energy intake. As mentioned in section 4.1, Raben et al. (2003) found that, when energy density was kept constant, varying the macronutrient content of meals did not affect satiety. The effect of energy density on satiety is discussed further in section 4.8.

Overall, it appears that when energy density is controlled, fat does not have a lesser effect on satiety than other macronutrients do. However, for free-living people whose conditions are not controlled, high-fat foods and diets will often also have a higher energy density than high-carbohydrate or protein foods. So, in practice, fat content and energy density are closely linked. The palatability of high-fat foods/diets is also a factor that could contribute to their over-consumption.

The chain length of fatty acids in a triaclyglyceride (fat) affects how they are absorbed and processed by the body (French 2004), and this and their degree of unsaturation (e.g. MUFA vs. PUFA) have also been suggested to influence satiety. However, where effects have been shown, they have generally been small or occur only on extreme diets (Stubbs & Harbron 1996; van Wymelbeke et al. 1998; French et al. 2000; Lawton et al. 2000; Krotkiewski 2001) and some studies have failed to demonstrate an effect (Kamphuis et al. 2001; Kovacs et al. 2001; Bendixen et al. 2002; Alfenas & Mattes 2003; Flint et al. 2003; MacIntosh et al. 2003).

4.6 Liquids and satiety

Trends in sugar-sweetened soft drink consumption have increased alongside increases in obesity prevalence, particularly in the US, leading to a speculation that these drinks may be partly responsible for the obesity epidemic (Malik et al. 2006). It has been suggested that caloric soft drinks could lead to excess energy consumption because energy from liquids fails to trigger satiety compared with equivalent energy intakes from solid food (Mattes 2006). Drewnowski and Bellisle (2007) reviewed the literature on liquid energy and weight gain and found that studies comparing the effects of equivalent amounts of liquid or solid energy on satiety have yielded inconsistent results and do not consistently support the hypothesis that liquid calories go undetected by appetite control systems. However, they emphasised that the consumption patterns of caloric soft drinks (i.e. consumption when thirsty and with meals rather than as a source of fuel when hungry) and their relatively low cost might be more important in determining their relationship with excess energy consumption and obesity than their physiological consequences (Drewnowski & Bellisle 2007).

Sugar-sweetened drinks have been the focus of much research, but it is also important to consider drinks containing protein or fibre, which may have greater effects on satiety, although there are few studies on these to date.

With regard to drinks containing protein, 1% milk (3.4 g protein/100g) did not exert a stronger effect on satiety than orange juice (0.5 g protein/100 g) or cola (zero protein content) (DellaValle et al. 2004). However, a yogurt drink containing 17.1 g of protein was found to be more satiating than a dairy fruit drink containing 2.6 g of protein, or a fruit drink that contained no protein (all were matched for energy and volume) (Tsuchiya et al. 2006), indicating that there may need to be a threshold amount of protein present to have a significant effect on satiety. With regard to fibre in drinks, fibre added to a beverage increased its viscosity and produced higher satiety ratings (Mattes & Rothacker 2001), and addition of a fruit puree to a drink has been found to increase the drink's satiating power (Haber et al. 1977).

In addition to the effects of drinks on satiety, the effect of liquid foods (in particular, soup) has also been investigated. Rolls et al. (1990) compared the effects of three energy-matched preloads – tomato soup, melon and cheese with crackers – on subsequent energy intake. Soup was the most effective at reducing subsequent energy intake. Although the cheese with crackers had a higher energy density, the melon and tomato soup were matched for energy density. Therefore, soup appears to have some additional satiating qualities (Rolls et al. 1990).

Mattes (2005) tested the effect of soups on appetite ratings and energy intake with energy-matched preloads of comparable solid foods (chicken soup vs. chicken breast, peanut soup vs. roasted peanuts and apple soup vs. whole apples) and a beverage (apple juice). Appetite ratings were comparable for the solid and liquid foods, all of which were rated more satiating than the apple juice. However, energy intakes tended to be lower after the soups than after the solid foods. Thus, soups appear to have a particularly satiating effect, which may be due in part to their lower energy density. It has also been suggested that cognitive factors are important in the satiating effect of soups. The fact that soups are seen as part of a meal and consumed in response to hunger, compared with drinks, which are consumed to address thirst or to accompany food, may have important implications for the effect of soups on satiety and energy intake (Mattes 2005).

Overall, there is inconsistent evidence to suggest that energy from liquids is less satiating than energy from solids. However, the mode of consumption and the perspective a person has on the function of the liquid (i.e. as a food or as a drink) may modulate the effect that liquids have on eating behaviour.

4.7 Alcohol and satiety

Raben et al. (2003) compared the effects of protein, carbohydrate, fat and alcohol on satiety and subsequent energy intake. The energy density of the meals was matched, and no differences were seen in appetite ratings or in energy intakes at the following meal (Raben et al. 2003). This suggests that, under controlled conditions, alcohol can stimulate satiety and reduce energy intake in a similar way to the other macronutrients. However, when alcohol is tested in a more realistic context (i.e. alcohol or a non-alcoholic control given as a drink before or with a meal), it does not reduce energy intake and indeed appears to stimulate appetite (Westerterp-Plantenga & Verwegen 1999; Hetherington et al. 2001; Yeomans & Phillips 2002). These effects occur fairly rapidly after alcohol consumption and so are likely to be a result of the pharmacological rather than metabolic effects of alcohol (Yeomans 2004).

The mechanisms by which alcohol could increase energy intake are not clear. Yeomans (2004) suggests that alcohol could act both by enhancing the hedonic (pleasurable) effects of food and by inhibiting satiation. In those who consciously restrain their eating, alcohol could have a disinhibiting effect, although it appears that the belief that alcohol has been consumed, rather than actual consumption of alcohol, mediates this effect (Yeomans 2004). Overall, lack of satiety and the stimulation of appetite after alcohol consumption may lead to passive over-consumption of energy. For more information on the effects of alcohol on energy intake, see Foster and Marriot (2006).

4.8 Energy density and satiety

Energy density is the amount of energy in a given weight of food or drink (kJ/g, kcal/g). A number of studies have shown that, when subjects are allowed free access to a range of foods, they will consistently consume a similar weight of food each day, rather than a constant amount of energy (Rolls 2000). This means that the lower the energy content of the foods eaten (i.e. the less energy dense they are), the lower overall energy intake will be and vice versa.

The primary determinants of energy density are water and fat, and foods with the lowest energy density are those with the most water and least fat. Conversely, high-energy dense foods are typically high in fat and low in water. Fibre can also help reduce energy density (Drewnowski 1998). It is important to note that energy density tends to be proportional to palatability, which itself affects satiety (see section 5.1), and therefore it is important for studies investigating the effects of energy density to control for palatability (Drewnowski 1998).

Investigating the effects of energy density on satiation (i.e. how much energy is consumed in one sitting), Bell et al. (1998) tested the quantity of foods with high-, medium- or low-energy density that female subjects consumed over two days, using a crossover design, so that subjects tried all three dishes on separate occasions. The dishes were matched for fat content and palatability, and energy density was reduced by adding vegetables. The subjects ate a similar weight of the meal in each case, which resulted in a 30% reduction in energy intake from the meal with the low- vs. high-energy density, without any differences in ratings of hunger or fullness (Bell et al. 1998). This study showed that higher-energy density foods were less satiating, but did not provide further information about satiety or subsequent energy intake.

Rolls et al. (1998) tested the effects of energy density on satiety using a milk-based preload, which was diluted with varying amounts of water to change the energy density. The energy and nutrient contents of the preloads were the same, but the volume increased to decrease energy density. The preloads were followed by an ad libitum lunch and then dinner. Energy intake at lunch was reduced by 18% after the lowest, compared with the highest energy dense preload, and this reduction in energy was not compensated for by an increase in energy intake at dinner (Rolls et al. 1998). As these studies had reduced energy density by adding water, this group also tested the effect of drinking water with a meal compared with incorporating it into the meal itself. Water added to the meal (chicken soup compared with a casserole with less added water) increased satiety and reduced subsequent energy intake. However, the equivalent amount of water served as a drink with the meal did not have the same effect (Rolls et al. 1999).

Incorporating air into a food to increase its volume without increasing energy content has also been found to increase satiety in a study by Rolls et al. (2000). Subjects consumed yogurt-based milkshakes that had been whipped up to increase their volume by incorporating air. The milkshakes contained the same amounts of energy and were of relatively high, medium or low volume. Energy intake after the milkshakes was 12% lower following the one with the highest compared with the lowest volume, and satiety was increased by both the high- and medium-volume milkshakes compared with the one with the lowest volume (Rolls et al. 2000). Although this study did not strictly alter the energy density in kJ/g, as the weight of the milkshakes was unchanged by the addition of air, simply increasing the volume of the drink without any additional energy led to a reduction in subsequent energy intake.

A number of studies have shown that covertly altering the energy density of foods results in significant differences in energy intake. Stubbs et al. (1995a) gave subjects a HF, medium-fat or low-fat diet for 7 days. Energy density varied in proportion to the fat content of the diet. Subjects consumed significantly more energy on the HF diet, without any concurrent increase in satiety; in fact, there were no differences in appetite ratings between the diet conditions (Stubbs et al. 1995a). These findings were confirmed in a follow-up study that tested the same diets over a two-week period (Stubbs et al. 1995b).

Covert manipulation of a high-carbohydrate, low-fat ad libitum diet to be of high- or low-energy density was investigated over a 14-day period (Stubbs et al. 1998a). Subjects consumed significantly less energy and lost more bodyweight on the low-energy density diet than on the high-energy density diet, but were also significantly hungrier. In a further study by the same group, varying the energy density of medium-fat diets, energy density increased energy intake without altering appetite ratings (Stubbs et al. 1998b).

Leahy et al. (2008a) looked at the effect of reducing the energy density of a dish served at lunch on energy intake in children aged between 2 and 5 years. Higher- and lower-energy density versions were served along with other side dishes, and children were allowed to eat ad libitum. The lower-energy density version led to a decrease in energy intake from both the dish itself and the lunch overall (Leahy et al. 2008a). This finding was confirmed by a further study that also tested changes in portion size. For both portion sizes, reduction in energy density reduced energy intake from the dish and the lunch. As the energy density was reduced by adding vegetables, this also led to an increase in the children's vegetable intake (Leahy et al. 2008b). The effect of reducing energy density on children's energy intakes was then tested for breakfast, lunch and snacks over two days. It was found that children consumed a consistent weight of food over the two days, so that energy intake was reduced when energy density was lower (Leahy et al. 2008b). Although appetite measures were not taken in these studies, the amount the children could eat was not restricted, and, if it is assumed that they ate to fullness, this would mean that the lower-energy density foods were no less satiating than the higher-energy density versions.

When eating patterns and bodyweight are assessed in the longer term, energy density also appears to affect energy intake and bodyweight. Greene et al. (2006) followed up people who had participated in a weight management programme approximately two years previously, which promoted the consumption of low-energy dense foods to determine which dietary pattern aided weight maintenance. All subjects ate a similar amount of food, but those who maintained their bodyweight ate a lower-energy dense diet overall and ate smaller portions of high-energy dense food than those who put on weight (Greene et al. 2006). No data were collected on satiety, but, again, if it is assumed that subjects ate until they were full rather than restraining their consumption, this suggests that satiety was not reduced by the lower-energy density diet.

A year-long trial compared two weight-loss diets, one where subjects were counselled to reduce fat intake (RF) and the other where they were counselled to reduce fat and to increase consumption of water-rich foods, particularly fruits and vegetables (RF + FV), reducing the energy density of the diet to a greater extent. Both groups were told to eat ad libitum within these recommendations. Both groups reduced their fat intake to a similar extent, but those on the RF+FV diet had a lower-energy density, consumed a larger weight of food and reported being less hungry than the RF group. Both groups lost weight, but the RF+FV group lost significantly more that the RF group (Ello-Martin et al. 2007).

A six-month intervention trial tested the effects of different levels of dietary advice, resulting in a reduction in energy density of the diet and weight loss. When subjects were categorised according to the reduction in energy density resulting from the intervention, it was found that both modest and large reductions in energy density were associated with weight loss (Ledikwe et al. 2007).

Saquib et al. (2008) looked at the effects of long-term reductions in energy density on bodyweight in a large group of overweight female breast cancer survivors over a four-year period. The intervention group members were counselled to lower their dietary energy density by reducing fat and increasing fruit, vegetable and fibre consumption and were given telephone support, cooking classes and newsletters to promote dietary change. The control group was given printed materials with US dietary guidelines and sent newsletters with health and nutrition information. The intervention group had a small but significant weight loss at one year. However, although the intervention group maintained a diet of lower-energy density compared with the control group, no differences in energy intake or weight loss were seen between the groups after four years. No information was collected on the effects of the diets on appetite (Saquib et al. 2008). Thus, it is not clear whether the apparent satiating effect of low-energy density diets in the short-term can help to maintain weight loss in the long-term, and further long-term studies are needed to investigate this.

A number of epidemiological studies (some cross-sectional and others prospective) have looked at whether there is an association between the energy density of the diet in different populations and bodyweight or weight gain. In cross-sectional studies, higher dietary energy density was found to be an independent predictor of obesity (Mendoza et al. 2007) and a risk factor for higher BMI or greater waist circumference (Howarth et al. 2006; Murakami et al. 2007).

In an eight-year prospective study of women (the Nurse's Health Study), increases in dietary energy density were associated with greater weight gain (Bes-Rastrolo et al. 2008) and, in another US study, lower energy density diets appeared to moderate weight gain (Savage et al. 2008). A cohort study of children in the UK found that higher energy density at 7, but not 5 years old was a risk factor for excess adiposity at 9 years old (Johnsone et al. 2008). McCaffrey et al. (2008) assessed the relationship between the energy density of children's diets at 6–8 years and change in adiposity to adolescence. Five different methods to assess energy density were used, but results from none of these methods were related to measures of body fatness in adolescence (McCaffrey et al. 2008). So there is a relatively (but not completely) consistent relationship between dietary energy density and weight gain or higher bodyweight. These studies did not collect any information on satiety, and, in free-living subjects who can eat as much as they like, increased satiety on a low-energy density diet could be a factor in the apparent reduction in energy intake. Overall, the energy density of foods and drinks appears to have a major impact on satiety and this may be more important than macronutrient composition in terms of enhancing satiety and helping to reduce the risk of weight gain.

Key points

  • • Studying the effects of specific variables within a food or drink, without affecting others, is inherently difficult, and, when considering the results from studies on the effects of foods and drinks on satiety, it is important to consider any confounding factors that could have had an effect.
  • • Energy from protein seems to have a greater effect on satiety than the other macronutrients.
  • • The effect of sugars on satiety has been investigated, and although they can stimulate satiety and reduce subsequent energy intake, this compensatory reduction in energy consumption may not be complete.
  • • Although it has been suggested that low-GI foods are more satiating than high-GI foods, current evidence shows that, when other factors such as fibre content are controlled, GI does not have a significant effect on satiety.
  • • Fat invokes a satiety response, but, in typical food compositions, it is associated with high energy density and palatability, so that it is readily over-consumed, leading to higher energy intakes.
  • • Some types of fibre can promote satiety, but this is highly dependent on the dose and especially on the type of fibre. Viscous fibres, such as petin, or novel gelling fibres, usually alginates, appear to be the most satiating.
  • • There has been debate surrounding whether intense sweeteners increase, decrease or have no effect on satiety, which is complicated by the different foods or drinks that contain intense sweeteners and the different types of sweetener used. Overall, evidence suggests that intense sweeteners do not increase or decrease satiety, although the reduction in energy intake associated with their use is not compensated for by a subsequent increase in energy intake.
  • • Energy from drinks has been suggested to be less satiating than that from solid foods. However, evidence to support this is conflicting. Studies using soups, which tend to be highly satiating, indicate that observed effects are not directly related to liquid vs. solid composition.
  • • Alcohol appears to stimulate appetite in the short-term and may lead to passive over-consumption of energy
  • • Energy density is important in the satiety response and may be more important than macronutrient content. Lower-energy density foods appear to be more satiating than higher-energy density foods.

5. The effect of external factors on satiation and satiety

Despite the internal mechanisms that exist to control appetite, most people know from experience that, in certain circumstances, people eat when they are sated and refrain from eating when hungry. Therefore, there are external factors that may affect or act independently of our responsiveness to internal signals of satiation and satiety that are important to take into account. Some of these are discussed in this section.

5.1 Palatability

Palatability refers to the pleasurable experience when consuming food. It is somewhat difficult to define, as it is not an inherent characteristic of a food, but a person's evaluation of his or her experience of eating food under particular circumstances (Yeomans 1998). Palatability is generally measured by using one of the rating scales described in section 3. Palatable stimuli act on hedonic pathways in the brain, stimulating a positive emotional response and increasing the drive to consume more (Berthoud 2007).

Increasing the palatability of food by, for example, adding fat, increases appetite, meal size, meal duration and eating rate (Yeomans 1998). Hence, if composition is not controlled, the most palatable foods tend to be the least satiating and the least palatable foods the most satiating (Drewnowski 1998). However, there is a strong relationship between palatability and energy density, in that higher-energy dense foods tend to be the most palatable and vice versa (Drewnowski 1998), perhaps reflecting innate preferences for energy-rich foods (Berthoud 2007). However, the relationship between energy density and palatability is not fixed, and the association between the two can be (experimentally) manipulated by altering the taste and appearance of foods while keeping energy density constant.

A study by Bobroff and Kissileff (1986) found that adding cumin to yoghurt reduced palatability and intake. Yeomans et al. (1997) added oregano to a tomato sauce served with pasta, which made the dish more palatable, and compared this with the same sauce without oregano, which was rated as bland. Hunger increased in the early stages of eating the dish with oregano and then declined, while it fell throughout consumption of the dish without the herb added. Subjects also ate more quickly when given the more palatable dish (Yeomans et al. 1997).

De Graaf et al. (1999) investigated the effects of reducing the palatability of a soup on satiation and satiety by adding increasing amounts of citric acid. Palatability was found to be inversely related to the amount of citric acid in the soup. Satiation was measured by allowing subjects to eat the soup ad libitum, and satiety was measured by giving a fixed amount and then measuring appetite. When the soup was offered ad libitum, the amounts consumed were inversely related to the palatability. No effect was seen on appetite ratings after the standard portions of soup, indicating that, in this case, palatability affected satiation but not satiety (De Graaf et al. 1999).

Poortvliet et al. (2007) attempted to dissociate the inverse relationship between palatability and satiety by designing a healthy meal that was both palatable and satiating and comparing this with a palatable control meal that was more energy dense. The healthy meal was high in fibre and protein and low in energy density, and, as energy was kept constant across the two meals, the serving size for the healthy meal was greater than the control (400 g vs. 185 g). Hunger, prospective consumption (how much subjects thought they could eat) and desire to eat were significantly reduced by the healthy meal compared with the control, and subsequent energy intake from a dessert after the meal was significantly lower (Poortvliet et al. 2007). Thus, although palatable energy dense foods have been shown to increase energy intake and reduce satiety, it appears to be possible to design lower energy density foods and diets that are both palatable and satiating.

It should be noted that, although the ‘liking’ of palatable foods may be important, the motivation to eat or the ‘wanting’ of foods should also be considered. These two factors may occur together, but evidence suggests that they may also operate separately (Mela 2006) and that, in the real world, ‘wanting’ may be more important than ‘liking’ (Castro & Plunkett 2000).

5.2 Variety

A number of studies have shown that the greater the choice of foods offered at a single eating occasion, the more people will eat (see Rolls 1984). This effect appears to be mediated by a phenomenon known as sensory-specific satiety (SSS), whereby the desire to eat a food that has already been tasted is significantly reduced compared with one that has not. This was initially demonstrated by allowing subjects to taste and rate the palatability of small portions of 8 or 9 foods, then offering an ad libitum meal containing one of these foods. After the meal, the foods that had been rated initially were then rated again, and the palatability of the food eaten at the test meal was found to have significantly decreased compared with the ones that had not (Rolls 1984). Hetherington et al. (1989) compared the palatability ratings of the foods after the test meal 2, 20, 40 and 60 minutes later and found that the decrease observed in the tasted vs. the non-tasted foods was most pronounced at 2 minutes after consumption. For the most palatable food, the palatability recovered somewhat after 60 minutes (Hetherington et al. 1989). The rapid nature of this response suggests that SSS occurs because of the sensory properties of food, rather than effects associated with digestion and absorption.

There also appear to be age-related differences in SSS. Rolls and McDermott (1991) tested SSS responses in adolescents, young adults, older adults and elderly subjects and found that the response was most pronounced in the adolescent subjects and decreased with age, with the elderly group showing the lowest response (Rolls & McDermott 1991).

It is not clear whether consuming an increased variety of foods in the longer term would lead to increased energy consumption, but it has been shown that consuming some foods (e.g. meat, shellfish) repeatedly for a period of time reduces their acceptance, although, for staple foods like bread and highly palatable items like desserts, this does not appear to be the case (Rolls 1984).

Overall, it appears that restricting the variety of foods eaten can stimulate satiety via SSS, and, conversely, increasing the variety of foods available could stimulate energy intake by avoiding SSS.

5.3 Portion size

In the US, portion sizes of many foods have been increasing since the 1970s (Young & Nestle 2002). There does not appear to have been a consistent increase or decrease in portion sizes in the UK, and, although the variety of portion sizes available has increased, standard portion sizes appear to have remained relatively constant (FSA 2008). If the amount eaten was determined only by internal satiation and satiety mechanisms, the portion sizes of foods served should not affect energy intake. However, it appears that, when presented with a larger portion, most people will consume more food (Ello-Martin et al. 2005).

Very young children are a notable exception to this pattern. When children 3 years old and younger were served increasing portions of macaroni and cheese, their energy intake remained constant. However, when the same portions were served to 5-year-old children, they significantly increased their energy intake as the portion sizes became larger (Rolls et al. 2000). This effect was also seen in 4-year-olds, who ate 25% more when they were served a portion double the size appropriate for their age (Fisher et al. 2003).

A number of experiments have tested the effect of increasing portion sizes on adults' energy intake in the short-term. These have used a number of different foods, including macaroni and cheese, sandwiches, crisps, pasta, salad and drinks, and consistently found that increasing the portion size leads to an increased energy intake. The effects on satiety vary, with two studies finding that satiety increased with bigger portion sizes, while the others found no differences in appetite ratings between larger and smaller portions sizes, despite a significant increase in energy intake when consuming the larger portion size (Rolls et al. 2002, 2004a, 2004b; Diliberti et al. 2004; Kral et al. 2004; Flood et al. 2006).

Further studies have investigated this effect over a longer period to assess whether increased energy intake from larger portion sizes is compensated for by a subsequent reduction in energy intake. Studies conducted over 2 days, 11 days and 2 months found that subjects continued to eat more for the period of the study when portion sizes were larger than when they were smaller, without reducing energy intakes to compensate for the additional energy consumed (Rolls et al. 2006, 2007; Jeffery et al. 2007).

Generally, greater portion size appears to increase energy intake. Some studies have found that this also increases satiety, while others have found no difference. So it appears that the ability of internal appetite controls to detect extra energy intake is limited, especially because no compensatory reduction in energy intake is typically observed. The fact that people eat more when given larger portion sizes also indicates that, for many people, the presence of food acts as a stimulus for (continued) consumption and that, within the time frame of a meal, this positive stimulation can be more powerful than the physiological signals that inhibit eating. It is interesting to note that very young children (<3 years) eat a consistent amount, regardless of portion size, and it may be that, at this age, children are more sensitive to internal signals of satiety (Rolls et al. 2000). There is some evidence that parental behaviour, such as pressuring children to finish their food, may have a negative effect on children's ability to control their energy intake (Savage et al. 2007).

It should also be noted that, within the studies mentioned above, there was an inter-individual variation in the response to portion size, and it is likely that individuals vary in their sensitivity to the effects of portion size on energy intake.

5.4 Sleep

Epidemiological studies have linked a chronic lack of sleep to obesity (Magee et al. 2008), and it has been suggested that restricted sleep has an impact on the appetite hormones leptin and ghrelin (Knutson 2007). As outlined in section 2.2.1, ghrelin, the so-called ‘hunger hormone’ secreted from the gut, appears to stimulate appetite and is suppressed after eating. Leptin is primarily produced by the adipose tissue in proportion to fat stores. A fall in leptin stimulates appetite, while an increase has an inhibitory effect (see section 2.2.2). Both hormones operate according to a circadian rhythm whereby they reach a peak level during sleep (Yildiz et al. 2004). Sleep restriction has been found to reduce leptin and increase ghrelin levels and to increase appetite (Spiegel et al. 2004). The association between reduced sleep and a reduction in leptin and increase in ghrelin was also observed in the Wisconsin Sleep Cohort study, although, in this case, no measures of appetite were taken (Taheri et al. 2004). The importance of these findings requires further study, as there may be other aspects of short sleep duration, apart from its potential effect on appetite, that could increase the risk of obesity, for example, having more time available to eat, reductions in physical activity when tired, or more unhealthy food choices (Knutson 2007).

5.5 Physical activity

Physical activity generally affects the body by increasing fitness, changing body composition by increasing lean body mass and increasing energy expenditure (see Miles 2007). The effects of physical activity on appetite and satiety have been investigated. For example, a number of studies have tested the effects of an acute bout of physical activity on appetite. These studies have consistently found that, even when large energy deficits are induced, this does not result in a compensatory increase in hunger and energy intake (see Blundell et al. 2003). This contrasts with studies that have induced negative energy balance by reducing energy intake, resulting in increased appetite and energy intake (Lawton et al. 1993; Green et al. 1994; Delargy et al. 1995; Hubert et al. 1998). Intense physical activity actually appears to suppress hunger for short periods (Thompson et al. 1988; Kissileff et al. 1990; King et al. 1994, 1996; Westerterp-Plantenga et al. 1997). This phenomenon, known as physical activity-induced anorexia, is not seen in moderate- or low-intensity physical activity and may be a result of the redistribution of blood flow away from the gut to the muscles (Blundell et al. 2003).

The effects of a physical activity-induced energy deficit in the longer-term (over a number of weeks) have been investigated in lean men and women, and it appears that subjects can tolerate this energy deficit without a compensatory increase in energy intake for approximately two weeks. After this, energy intake begins to increase, but, initially, compensation is incomplete, accounting for approximately 30% of the deficit on average after 16 days. There were inter-individual differences in how much subjects adjusted their energy intake to account for increases in physical activity, and it is not clear what the reasons for this were (Wyebrow et al. 2008).

Lim and Lee (1994) investigated the impact of five months of military training, while eating ad libitum. They found that lean body mass remained constant, but body fat was reduced progressively throughout the training. Consequently, subjects who were initially fatter lost most weight, while those who were initially lean maintained their bodyweight (Lim & Lee 1994). As the function of stored fat is to supply the body with energy when demands are not met by energy intakes and/or the need for physical activity is high, it may be that excess adipose tissue buffers the effect of an energy deficit and that a compensatory increase in energy intake occurs when lean body mass is threatened (Blundell et al. 2003). The evidence from intervention studies using physical activity to promote weight loss or prevent weight gain is mixed. Thus, physical activity-induced negative energy balance may be reduced over time until the individual compensates completely for the additional energy expenditure. However, there may also be issues with compliance in these studies (see Miles 2007).

Stubbs et al. (2004) investigated the effects of a reduction in physical activity by imposing a sedentary routine on lean men for nine days and comparing their energy intakes and appetite ratings with those when they had a moderately active routine. The reduction in physical activity caused a significant reduction in energy expenditure compared with the active routine, but this did not lead to reductions in appetite or energy intake, resulting in positive energy balance and weight gain on the sedentary regime (Stubbs et al. 2004).

Thus, it appears that satiety pathways may not always detect changes in energy balance caused by physical activity. This is the case both when physical activity is increased, which can lead to substantial energy deficits without corresponding reductions in satiety and increases in energy intake, and when it is reduced, where the reduction in energy expenditure does not appear to increase satiety or reduce energy intake. Although large energy deficits due to physical activity cannot be maintained in the long-term, physical activity can aid short-term weight control and help maintain a healthy weight.

5.6 Television viewing and other distractions

Some studies have reported that children (Bellissimo et al. 2007; Temple et al. 2007; Manios et al. 2008), adolescents (Van den Bulck & Van Mierlo 2004) and adults (Stroebele & de Castro 2004) increase their energy intake when eating while watching television, compared with eating undistracted.

Some of these studies monitored appetite as well as energy intake. Bellissimo et al. (2007) monitored the appetite ratings and energy intakes at lunch of boys between 9 and 14 years old, with and without watching television. They also tested the ability of the boys to compensate for a glucose-containing preload at the subsequent lunch, again, with and without television watching. Watching television significantly increased the boys' energy intake at lunch. When not watching television, the boys reduced their energy intake at lunch to compensate for the glucose preload, but did not do this as effectively when watching television at lunch time. No significant differences were seen in appetite ratings, despite the differences in energy intake. The authors concluded that television viewing had made the boys less sensitive to internal signals of satiation and satiety (Bellissimo et al. 2007).

Temple et al. (2007) looked at the effect of watching television on children's motivation to eat and their energy intake. Firstly, children were given a computer-based task that had to be completed to earn portions of food. The group that watched television while doing this spent more time doing the tasks and ate more food than the control group who were not watching television. The second experiment measured the effect on energy intake of a continuous segment of television, which was compared with showing repeated short segments and to no television. The group that watched a continuous segment of television consumed more energy and spent longer eating than the other groups. Based on these results, the authors suggested that television increased motivation to eat and that the attention needed to watch a continuous television programme distracted children from their habitual eating patterns (Temple et al. 2007).

In a review of cognitive influences on eating behaviour, Higgs (2008) concluded that the memory of a previous meal reduces subsequent snack intake and that, if the formation of this memory is disrupted, subsequent energy intake is increased. This effect can be seen in amnesiacs who eat multiple meals because they do not remember the previous eating occasion (Higgs 2008). In a subsequent study, Higgs et al. (2009) found that watching television while eating reduced the vividness of the memory of the meal, compared with a eating a meal without distraction. Snack intake after the meal with television was higher than that without, suggesting that the disruption of the memory of the meal increased energy intake (Higgs et al. 2009). It is possible that this effect could apply to any form of distraction that affects the memory of eating.

The idea that distractions while eating can influence energy intake was explored by Bellisle and Dalix (2001) in women who cognitively restrain their eating (the effect of dietary restraint is discussed in section 3.3.2). They tested the effects of four conditions on women's energy intake: eating alone, eating alone while listening to a recording that encouraged them to pay attention to their food, eating alone while listening to a recorded detective story, and eating in a group. The first two conditions were considered as non-distracting and the second two as distracting. In both the ‘distracting’ conditions, the women consumed more energy than in the ‘non-distracting’ conditions, but this was not reflected in differences in appetite ratings (Bellisle & Dalix 2001). The results from this study may reflect a loss of cognitive restraint in these women when distracted, or that distraction inhibited the effects of internal signals of satiation and satiety. Listening to music while eating has also been shown to increase energy intake compared with eating without music (Stroebele & de Castro 2006).

Overall, there is some evidence that distractions when eating (commonly television) can make people less sensitive to internal appetite controls and lead to increased energy intake. The relationship between television viewing and weight gain is complex, as watching television introduces more sedentary time as well as the opportunity to eat more (Crawford et al. 1999). However, a recent review suggests that part of the association that appears to exist between obesity and time spent watching television may be mediated by increased consumption of food and drink while watching television (Cleland et al. 2008).

5.7 Social situations

It has been shown that, typically, people eat significantly more (on average 44%) when with other people than they do alone (De Castro & De Castro 1989). In addition, the satiating power of these meals per unit of energy is lower in those eaten in a group than those eaten alone. In fact, there appears to be a direct relationship between the number of people present at a meal and the amount that the diners eat, so that the more people present at a meal, the more energy is consumed (De Castro & De Castro 1989). This may be because, in the presence of people eating large amounts, others will imitate this and consume more. It has also been suggested that the emotional effect of eating with others stimulates energy intake, that social interactions while eating might facilitate disinhibition (i.e. relaxing any cognitive restraint when eating) and finally, that eating in groups might simply mean that people will eat for longer and thus eat more (De Castro 1994; Feunekes et al. 1995).

De Castro (1994) looked at the effect of different dining companions on energy intake: spouses, family, friends, co-workers and others. In all cases, meals eaten with other people were larger and longer in duration than those eaten alone. However, there were some differences depending on the eating companions. Meals eaten with spouses and family were larger and eaten faster than those with other companions, while meals eaten with friends were also larger but of longer duration. The author suggested that, in general, the social interaction of eating with others prolongs meal times, thus increasing energy intake. Eating with family and friends may have increased this effect by relaxing diners and allowing disinhibition of restraint, making energy intake even greater (De Castro 1994).

Kristensen et al. (2002) performed a qualitative study in Denmark that investigated 20 people's perceptions of ‘proper satiety’ in different social situations, in order to gather information on the social contexts and cultural norms surrounding satiety. There was a variation in the subjects' preferences regarding satiety. Some preferred a light feeling of satiety, as experienced when eating vegetables, and associated being full after high-fat foods with feeling sluggish and lethargic; while others preferred the feeling of being really full and felt unsatisfied without this. Many subjects reported enjoying eating to excess with friends, and enjoyed the feeling of being sated and lazy but said that they avoided this with those who they did not know well. Conversely, those who described themselves as having a weight problem tended to limit their intake when eating with others and were more likely to eat in excess when alone. Female subjects reported feelings of guilt surrounding excessive eating and the subsequent feelings of satiety, whereas males generally did not. Many subjects felt the need to control food intake, with the aim of limiting satiety, if at work or working at home, in order to be more alert, and meals in this context were viewed as more functional than pleasurable. However, meals eaten in one's spare time were enjoyed more, and it was seen as more acceptable to eat more and to feel more sated after eating (Kristensen et al. 2002). Although this study must be viewed in context of its location and small size, it highlights the fact that, as well as internal mechanisms of satiation and satiety, there will be expectations about satiety that can change according to the social context of the situation, and these will also influence eating behaviour.

Key points

  • • Palatability tends to reduce satiation and satiety and leads to increased energy intake. Palatable foods are often energy dense, which may drive this relationship, but the relationship can also be manipulated by using different ingredients and flavours.
  • • Offering a variety of foods seems to decrease satiety and increase energy intake within a meal. ‘Sensory-specific satiety’ describes the process of reduced palatability and increased satiety for a particular food when it is consumed repeatedly, and it is this process that drives increased consumption if alternative foods are offered.
  • • Most people consume more energy when offered larger portions of food, and satiety does not appear to be increased after this greater energy intake.
  • • Children younger than four years do not consume more when given a larger portion size, which may be a result of a greater sensitivity to internal satiation and satiety signals.
  • • Lack of sleep has been associated with increased bodyweight. Some studies have shown that sleep deprivation affects hunger and satiety signals in a way that would be expected to reduce satiety, but this requires further investigation.
  • • In the short-term, large energy deficits can be induced by physical activity, without a corresponding reduction in satiety. However, this cannot be maintained indefinitely, and the negative energy balance is gradually compensated for by an increase in energy intake over the longer-term. Nevertheless, there are inter-individual differences in the extent to which the energy deficit is compensated for.
  • • People tend to eat more and be less responsive to satiety signals when distracted by television or other means, and television watching has been associated with an increased risk of obesity. This association is complicated by the sedentary nature of television watching, which could also contribute to obesity risk.
  • • When eating with others, people consume significantly more energy than when eating alone. This may be the result of a number of factors, such as longer duration of eating and to a relaxation of dietary restraint.

6. Satiation, satiety and weight control

Levels of obesity in the UK have tripled since the 1980s (National Audit Office 2001), and this trend is predicted to increase so it is estimated that over half of UK adults will be obese by 2050 (Foresight 2007). This rapid increase suggests that it is environmental rather than genetic changes that have lead to the increased risk of weight gain. However, within this current ‘obesogenic’ environment, certain individuals appear to be at greater risk of developing obesity than others. The reasons for this are complex and involve physiological, psychological and environmental factors (Foresight 2007). Appetite control, including the development of satiation and satiety, is one of the variables that are important to consider in weight management.

This section evaluates some of the factors that may affect either the development of satiation and satiety or an individuals' responsiveness to these internal signals, and how these may modulate the risk of obesity. There are a number of levels at which differences in satiation and satiety could increase the risk of obesity. Genetic differences, for example, may impact on components of satiation and satiety signalling pathways. There may also be physiological differences in the way that satiation and satiety develop or are linked into reward and learning systems that influence eating behaviour. Behavioural, neurological and psychological factors may also affect how much internal appetite signals control energy intake compared with the hedonic experience of eating or external influences on energy intake.

6.1 Obesity genes and satiety

In relatively rare cases, obesity may be linked to a single mutation in a gene that renders its product non-functional. All obesity genes that have been characterised appear to cause obesity by disrupting satiety (O'Rahilly & Farooqi 2008). For example, mutations in the leptin gene cause extreme over-eating and obesity in early childhood. This can be dramatically reversed by administration of leptin (Gibson et al. 2004). Mutations in the melanocortin 4 receptor (MC4R) gene are the most common mutation linked to human obesity (Farooqi et al. 2003). As outlined in section 2.3.1, MC4R is expressed in appetite centres in the brain and conveys signals of satiety, resulting in a reduction in energy intake. Humans have two copies of every gene, and mutation of both copies of the MC4R gene results in over-eating and severe obesity. If only one copy of the gene is affected, these symptoms are still present but are less severe (Farooqi et al. 2003).

Gene mutations, such as those described above, account for a very small proportion of cases of human obesity. However, it has been estimated that a person's genetic profile may account for 45% to 75% of their variation in BMI (Farooqi & O'Rahilly 2007), indicating that, within the current ‘obesogenic’ environment, genetic profile affects an individual's risk of obesity. Although all humans have the same set of genes, variations known as polymorphisms give rise to interindividual differences between people. Polymorphisms in over 100 genes have been linked to obesity (Rankinen et al. 2005) and, although their contribution to increasing or decreasing the risk of obesity is relatively small, it is possible that they are part of a phenotype (the physical expression of a genetic profile) that increases the likelihood of weight gain.

Many genes that have been linked to obesity are involved in the control of appetite, and a recent update of 15 new genetic locations linked with BMI found that the roles of all of the proteins the genes code for could be traced to appetite centres in the hypothalamus (Hofker & Wijmenga 2009). One example is the FTO gene, of which certain polymorphisms were found to increase or decrease the risk of obesity (Frayling et al. 2007). Polymorphisms in the FTO gene that increase the risk of obesity have recently been linked with increased energy intake in adults (Haupt et al. 2008) and children (Cecil et al. 2008; Wardle et al. 2009). Wardle et al. suggested that the low risk form of the FTO gene was protective against this increased energy intake by promoting responsiveness to internal signals of satiety (Wardle et al. 2009). Thus, overall, the evidence strongly implies that the genetic profile can influence the risk of obesity via components of appetite control systems and pathways involved in hedonic responsiveness.

6.2 Physiological differences in satiation and satiety responses in obese people

In addition to genetic differences that may influence appetite, there may also be differences in the physiological signals of satiation and satiety that influence risk of obesity.

As discussed in section 2.1.1, gastric distension is involved in satiation. Some studies have found that obese subjects have greater gastric capacities (Geliebter 1988; Kim et al. 2001), which could result in a greater energy intake before gastric distension-stimulated satiation occurs. However, this may be related to binge eating and not strictly to obesity itself, and, although the likelihood of binge eating increases with the degree of obesity, the gastric capacity of normal weight bulimic patients has been found to be larger than that of some obese subjects (Geliebter & Hashim 2001). The gastric capacity of obese subjects can be reduced after weight loss, and it is possible that this could help restore a stronger satiation response (Geliebter et al. 1996).

There may also be differences in satiety signalling in obese people. Ghrelin increases hunger and is suppressed after energy intake (see section 2.2.1). Ghrelin levels are significantly lower in obese than in lean subjects (English et al. 2002). In addition, the post-prandial suppression of ghrelin that occurs in lean subjects is not observed in obese people. This may be because levels of ghrelin have already been maximally suppressed because of excess fat storage (English et al. 2002). Morbidly obese subjects appear to have an altered pattern of hunger without the pre-prandial rise and post-prandial fall in hunger seen in lean individuals. However, it has been found that normal hunger profiles can be restored in these morbidly obese subjects by administering exogenous ghrelin (Huda et al. 2009). Although suppression of ghrelin in obesity may be an adaptive response to excess energy storage, any effects on reducing hunger do not appear to be reflected in reductions in energy intake and bodyweight. Indeed, it has been suggested that disruption of hunger may result in continuous ‘grazing’ rather than defined meals (Huda et al. 2009).

Gut hormones that signal satiety are also altered in the obese. Release of PYY, which is secreted from the gut after feeding and induces satiety (see section 2.2.1), is lower in obese than in lean subjects (Batterham et al. 2003a), and this is associated with reduced satiety (le Roux et al. 2006). Although a lack of PYY could potentially contribute to an increased risk of obesity, it is also possible that obesity itself causes a reduction in PYY release. Release of GLP-1, another gut hormone involved in satiety (section 2.2.1), has also been found to be attenuated in the obese (Ranganath et al. 1996). When GLP-1 release was compared in lean, obese and obese subjects who had significantly reduced their bodyweight, the obese subjects who had reduced their bodyweight released GLP-1 at a level between that of the lean and obese subjects (Verdich et al. 2001). This indicates that weight loss improves GLP-1 secretion and that impaired GLP-1 secretion may be a consequence of weight gain.

PYY and GLP-1 are raised significantly after gastric bypass surgery to treat obesity because of the drastic change in the structure of the gut, and it is thought that this contributes to the weight loss seen after surgery (le Roux et al. 2007). In addition, a number of gut hormones are being tested in clinical trials as obesity treatments (Field et al. 2008). Therefore, changes in gut hormones that have been implicated in increasing the risk of obesity may also be useful in treating it.

Thus, some aspects of the physiological induction of satiety appear to be altered in obese people. As these are only parts of the complex physiological control of appetite, it is difficult to determine how much they may be involved in increasing the risk of weight gain or inhibiting weight loss. However, they may be important when considering the relationship between satiation, satiety and obesity.

6.3 Behavioural differences in the response to satiation and satiety in obesity

Previous sections in this chapter have outlined genetic and physiological differences that may affect satiation and satiety and thus bodyweight. There may also be behavioural differences that affect an individual's responsiveness to internal signals of satiation and satiety.

The way people eat may encourage excess energy intake and weight gain. For example, a number of studies have found associations between bodyweight and the speed of eating (Barkeling et al. 1992; Sasaki et al. 2003; Otsuka et al. 2006; Maruyama et al. 2008). Laessle et al. (2007) studied the eating behaviour of lean and obese individuals and found that obese subjects ate faster, took bigger spoonfuls and had a greater overall energy intake than lean subjects. Differences in the rate of eating may encourage excess energy intake before internal signals of satiation can take effect, predisposing those with faster eating rates to positive energy balance and weight gain. Equally, as obese people have greater energy requirements, it is possible that faster eating is related to consuming larger meals in order to maintain energy balance and is a consequence, rather than a cause, of obesity.

High consumption of high-fat foods is a risk factor for obesity (e.g.Bray et al. 2004). An interesting link between preference for fat and behaviour has been described by Blundell et al. (2005). When characterising obese subjects with habitually high-fat diets (not all those with a habitually high proportion of fat in the diet were obese), it was found that their eating behaviour seemed to predispose them to weight gain. Their satiety response to fatty meals was weaker, and satiety did not reduce their preference for fatty foods. They also had greater hedonic responses to palatable foods and to eating in general, and reported greater feelings of hunger and were more prone to disinhibition (the loss of control over eating) (Blundell et al. 2005). This cluster of behavioural characteristics appears to predispose these individuals to gain weight in the current ‘obesogenic’ environment.

With regard to food preferences, it should be noted that this concept can be divided into two components: ‘liking’ and ‘wanting’. These describe the difference between enjoying the taste of a food (liking) and the desire to consume it (wanting), and it is possible for these to operate separately. The relationship between obesity and increased ‘liking’ of foods is not consistent, and it has been suggested that it is an increased motivation to eat (wanting), not necessarily accompanied by an increased liking of foods, that could predispose individuals to obesity (Mela 2006).

Another aspect of research into the relationship between eating behaviour and weight control is the ‘externality theory’ of obesity, which suggests that susceptibility to external cues such as time, presence of food and situation is greater in the obese (Schachter 1968). However, the relationship between the external environment, eating behaviour and obesity is complex and difficult to measure, and results from studies investigating this have been inconsistent (Mela 1996). It has been suggested that susceptibility to over-consuming palatable foods is the external factor that most consistently accounts for differences between lean and obese subjects (Spitzer & Rodin 1981). This susceptibility to external influences may mean that internal signals of satiation and satiety have less impact on energy intake in obese individuals.

Carnell and Wardle (2008) investigated the effect of responsiveness to internal satiety signals and to external food cues on BMI and waist circumference in 2- to 3-year-old and 8- to 11-year-old children. In both age groups, lower satiety responsiveness and higher food cue responsiveness were associated with greater BMI and waist circumference (Carnell & Wardle 2008). The authors highlighted the fact that it is important to distinguish whether low satiety responsiveness and susceptibility to external food cues operate separately or are, in effect, two sides of the same coin. In this study, these two factors were correlated but also contributed independently to predicting adiposity, suggesting that they are related but separate (Carnell & Wardle 2008). Further studies to determine the mechanisms behind these phenomena are needed in order to disentangle this relationship.

In addition to the possibility that susceptibility to external cues may influence eating behaviour, it is also possible that variations in a person's sensitivity to the internal hedonic response to foods may also play a role. The role of ‘sensitivity to reward’ (STR), (i.e. the capacity to experience pleasure) has been investigated with regards to the risk of weight gain. There are two possible routes by which STR could affect energy intake and risk of obesity. Heightened STR could stimulate excess consumption of palatable foods because of their increased reward value. Alternatively, reduced STR could increase consumption of palatable food in order to gain an optimal level of hedonic stimulation (Finlayson et al. 2007). There is some evidence that both may be involved in increasing the risk of obesity.

Franken and Muris (2005) found a direct relationship between STR and BMI. However, Davis et al. (2004) measured STR in lean, overweight and obese subjects and found a U-shaped relationship, with overweight subjects having greater STR than both lean and obese subjects (Davis et al. 2004). The possibility of a lower capacity to experience reward in obese people has also been observed in brain-imaging studies. The dopamine D2 receptor, which is involved in hedonic pathways in the brain, has been found to be decreased in obese compared with lean subjects, in proportion to BMI (Wang et al. 2001). Thus, it appears that the relationship between STR and obesity is complex. It has been suggested that hedonically driven over-consumption of palatable foods may eventually lead to down-regulation of the dopamine system in the brain, leading to a reduction in the capacity to experience reward from food. This would mean that high STR could be a risk factor for weight gain but not necessarily a characteristic of obese people (Blundell & Finlayson 2004).

Overall, it is emerging that variations in eating behaviour may be particularly important in determining the risk of obesity. Within the current environment, which is replete with food cues, it may be a person's overall sensitivity to positive stimuli to eat, rather than single specific defects in internal signals of satiation and satiety, that play the most important role in weight gain.

Key points

  • • Satiation and satiety are important in controlling energy intake and, therefore, should be taken into account when considering weight control and the risk of obesity.
  • • The majority of obesity genes that have been characterised affect the risk of obesity via effects on appetite control. In rare cases, single-gene mutations can cause severe obesity, but more commonly, there appear to be particular genetic profiles that affect the risk of obesity within the current ‘obesogenic’ environment.
  • • There are some physiological differences in the development of satiation and satiety that have been observed in obese people, including increased gastric capacity and differences in the production of gut hormones. However, it is not clear whether these are a cause or effect of obesity.
  • • Differences in eating behaviour affect an individual's responsiveness to internal signals of satiety, and these may predispose certain people to excessive energy intake and obesity.
  • • Knowledge about the relationship between satiation, satiety and eating behaviour may help when designing strategies to reduce or prevent obesity in the future.

7. Conclusions

Satiation, satiety and their effects on eating behaviour are important when considering energy intake, particularly with regard to weight control and obesity.

Satiation and satiety are typically measured by using a combination of energy intake and self-reported appetite ratings, and there are many factors that can potentially confound these measurements. For these confounders to be controlled, studies are generally performed in a laboratory setting and are relatively short-term (i.e. 6–12 hours per test condition). It is much more difficult to get reliable data about satiation and satiety in the longer term, outside a laboratory setting, because of problems in obtaining accurate self-reported dietary intake data. This raises issues as to how far laboratory data can be extrapolated to free-living people.

Energy density appears to be the major dietary factor influencing satiation and satiety, and there is a substantial body of evidence to suggest that low-energy density foods and diets promote satiation and satiety and may help to control weight. For this to be of benefit to people who wish to prevent weight gain or lose weight, practical guidance is needed on how to reduce dietary energy density. There are a variety of ways to reduce the energy density of the diet, for example, eating only small portions of fatty foods, increasing the water content of dishes, eating more fruit and vegetables and choosing higher-fibre foods. This could provide the flexibility to make this approach acceptable to consumers who want to control their weight.

However, regardless of how effective a dietary manipulation might be in enhancing satiation and satiety, the fact remains that people do not always respond well to these internal signals. Free-living individuals often eat when sated and sometimes refrain from eating when hungry. Therefore, an understanding of the factors other than satiation and satiety that affect eating behaviour, is essential. Evidence is emerging that inter-individual variations in sensitivity to internal appetite signals and responsiveness to the influence of external eating cues may underlie the differences in susceptibility to obesity that are seen in the current environment.

It is clear from the increasing worldwide prevalence of obesity that internal signals of satiation and satiety are not sufficient to restrain energy intake in many individuals. Systems to control energy balance in the body evolved under conditions where energy intake was directly related to energy expenditure and where the storage of excess energy as fat could enable humans to survive periods when food was not abundant. However, the current environment in much of the world, where food is plentiful and accessible and physical activity is largely unnecessary, means that, in susceptible people, these homeostatic controls are simply overwhelmed by positive stimuli to eat.

Fundamental changes in our environment and behaviour are needed to halt and reverse current trends in obesity prevalence. Knowledge about the effects of satiation and satiety on eating behaviour and how these are manipulated can contribute to reversal of current trends, provided that this knowledge is translated into practical advice that helps people control their energy intake and achieve a healthy bodyweight.

Acknowledgements

The Foundation wishes to thank the members of the Foundation's Scientific Advisory and Industrial Scientists Committees who have helped to shape the contents of this Briefing Paper, and Professor John Blundell, Professor Barbara Livingstone and Dr David Mela who reviewed the paper. The Foundation also wishes to thank Kellogg's for their financial support in writing the Briefing Paper.

Conflict of interest

The Foundation has received financial support from Kellogg's in the production of this Briefing Paper. However, the views expressed are independent and Kellogg's have not been involved in writing or shaping the contents of this paper.

Ancillary