Appetite control – biology or culture
Food intake is viewed by different groups of researchers as an exclusively biological or cultural phenomenon. On the one hand, for some theorists, eating is a way of getting food into the body to provide energy and nutrients for maintenance and growth. This implies a biological purpose; other features – such as opportunities for personal interaction or social discourse – are regarded as mere embellishments not related to the true purpose. In contrast, other groups of researchers, such as anthropologists or sociologists, construe eating as a means of fulfilling social obligations, embodying rituals and influencing personal relations1 and any physiological consequences are regarded as a side-effect. These opposing positions create a conflict which can be resolved by viewing biological and cultural determinism as incomplete explanations. Eating has its origins in biological obligations but the integrated patterns of eating are shaped and given values though the influence of culture. Part of the framework for understanding food intake is to see eating as a biocultural interaction (Figure 1). Feeding is a form of behaviour and therefore eating can be seen as a biobehavioural bridge through which feeding expresses biological needs in a social landscape.
Appetite control – key terms
Two fundamental questions about food intake in humans are: ‘what to eat?’ and ‘how much to eat?’. Two subsidiary questions are where and with whom to eat – neither of which is trivial – but these will not be the focus of this review. In addition, it should be recognized that food consumption (or energy intake) is 100% behaviour and therefore eating is a phenomenon that is subject to laws of behaviour rather than rules of physiology. Some key terms that are useful to define are appetite, hunger, satiation and satiety.
Appetite: there are two definitions in circulation:
covers the whole field of food intake, selection, motivation and preference;
refers specifically to qualitative aspects of eating, sensory aspects or responsiveness to environmental stimuli, which can be contrasted with the homeostatic view based on eating in response to physiological stimuli, energy deficit, etc.
a construct or intervening variable that connotes the drive to eat. Not directly measurable but can be inferred from objective conditions;
a conscious sensation reflecting a mental urge to eat. Can be traced to changes in physical sensations in parts of the body – stomach, limbs or head. In its strong form it may include feelings of lightheadedness, weakness or emptiness in the stomach. Hunger will be used in this sense throughout this review.
Satiation: process that leads to the termination of eating; therefore, controls meal size. Influenced by sensory qualities, palatability, energy density, density (weight and volume) and portion size. This is also known as intrameal satiety.
Satiety: process that leads to inhibition of further eating, decline in hunger, increase in fullness after a meal has finished. Influenced by total energy, macronutrient composition, types of fibres, carbohydrates, proteins and fats. This is also known as post-ingestive satiety or intermeal satiety.
Humans are omnivores
One of the most salient features of appetite in humans depends on the fact that humans are omnivores. Unlike herbivores or carnivores whose feeding habits are biologically programmed for restricted types of foods, omnivores have a much greater range of potentially edible items. One consequence of this has been to enable humans to colonize and exploit many different types of environments sustaining quite distinctive nutritional repertoires. It follows that, for humans, the type of food that is put into the mouth is not heavily programmed biologically but depends on the local culture, geography, climate, religion, ethnic principle and social forces. This means that the processes of food intake control have to be geared to a variety of dietary scenarios and the control mechanisms have to be sufficiently adaptable to deal with a huge range of food types. Although food is basically made up of fats, proteins and carbohydrates, it is put into the mouth in a large number of forms and associated with a multitude of tastes and textures. The behavioural act of putting a selected type of food into the mouth is a precursor of eating. This means that food choice depends on the environment. In certain environments rational food choice can be undermined by an environment in which nutritional value (and therefore biological value) of specific food items can be concealed or confused. This can easily happen in technologically advanced societies in which synthetic foods can be readily manufactured and which contain arbitrary and unlikely combinations of composition, textures and tastes. This can lead to quantitative and qualitative inappropriate eating habits.
Overconsumption, energy balance and the obesogenic environment
The fact that humans are omnivorous, together with the huge variety of manufactured foods, may be responsible for the epidemic of obesity. It is claimed that the epidemic is maintained (and possibly initiated) by an ‘obesogenic environment’ that encourages overconsumption.2,3 In turn, it has been argued that overconsumption is ‘passive’ and depends on the energy density of the diet.4 This overconsumption is permitted because of the asymmetry of the appetite control system in which excess food intake is readily allowed whereas underconsumption (biological deficit) is strongly resisted. In other words, overeating is easy (in an obesogenic environment) but undereating is difficult. Consequently, the role of food in excessive energy intake can be seen to be both economic and political but made possible by the biological asymmetry of the appetite system, and by the presence of a persistent biological drive.
It is widely considered that weight gain can be achieved only with a surfeit of energy intake (EI) over energy expenditure (EE), and this draws attention to the concept of energy balance. This concept is often depicted as set of kitchen scales with food on one pan and physical activity on the other; whichever has the greatest value causes either an energy surfeit or deficit. However, it has been recognized for some years that this simple mechanistic model gives a false representation.5 The balance mechanism is not a simple physical device but an active physiologically regulated system. This means that food intake not only influences the EI side of the equation, but also has an effect on EE.6 In turn, physical activity not only influences EE but also has an effect on EI.7 This picture has been referred to as ‘dynamic’ rather than ‘static’ energy balance8 and is consistent with new ways of viewing appetite control.9
The Satiety Cascade – episodic control of food intake
In the theoretical treatment of food intake control (in relation to body weight) it is conventional to refer to long- and short-term processes; or tonic and episodic. The tonic controllers change gradually over time and they reflect enduring and slowly changing mechanisms [such as fat-free mass (FFM), fat mass (FM) and resting metabolic rate (RMR)]. In contrast, the episodic controllers fluctuate markedly during the course of a day in relation to the episodes of eating; these mechanisms are considered to be responsible for controlling the size of meals and the period of inhibition over eating that follows a meal (post-prandial period). The ‘Satiety Cascade’ was developed to account for the episodic control of eating and to provide a framework for thinking about the different mechanisms involved: nutritional, physiological and psychological. The Satiety Cascade reflects the fact that eating itself is influenced by a number of factors including the sight, taste, texture and smell of food, thoughts and beliefs about food, as well as the actual nutritional composition and the environmental factors surrounding eating.10 The graphic in Figure 2 shows a model of the Satiety Cascade illustrating that the influence of overlapping variables such as sensory and cognitive processes are integrated with physiological consequences – all generated by the act of food consumption. Acting conjointly these factors influence the expression of eating behaviour and therefore determine, in part, the amount of food consumed and the willingness to eat.
The Satiety Cascade indicates that eating is strongly influenced by physiological processes induced by food consumption. Within the gastrointestinal tract a number of hormones and peptide neurotransmitters are secreted which influence the intensity and duration of satiety. Many peripherally administered peptides cause changes in food consumption. A number of these physiological signals (peptides released from the gastrointestinal tract following the ingestion of food) have now been identified. These include cholecystokinin (CCK), peptide tyrosine tyrosine (PYY), glucagon-like peptide-1 (GLP-1), oxyntomodulin, gastric inhibitory peptide and others. Experiments have revealed that their activity is related to the inhibition of eating (i.e. satiety). Hence, they can be referred to as satiety signals or satiety hormones. Indeed, there is currently much interest in developing analogues of satiety peptides as antiobesity pharmaceutical agents; one such drug is liraglutide (Victoza®, Novo Nordisk, Bagsvaerd, Denmark), which is a GLP-1 agonist.
It is often said by physiologists that a state of satiety is conferred by an intestinal tract full of food. This is almost certainly the case in every culture. The effect is mediated by the release of peptides from specialized cells that respond to ingested food (as noted above). However, not all peptide-signalling molecules are inhibitory. In recent years, the hormone peptide ghrelin has been identified whose action appears to stimulate hunger and eating. Therefore, signals from the gastrointestinal tract can be both excitatory and inhibitory. In addition, we have to remember that even powerful physiological signals can be over-ridden by conscious action. That is, people can volitionally ignore their own physiological signals and continue to eat in a state of satiety.
Satiation and satiety
Two fundamental components of the Satiety Cascade are satiation and satiety. Satiation represents the effect of those processes that occur whilst a person is in contact with food during the eating process itself. These processes influence the duration and then the termination of an episode of eating and therefore control meal size; along with the composition of food being eaten, these determine the amount of food energy put into the mouth. Following the termination of a meal (or other eating episode) the effects of ingested food exert a control over the processes of satiety which reflects the suppression of the motivation to eat and an inhibition of eating behaviour for a measurable period of time. Satiety is often best indicated by changes in subjective appetite sensations such as hunger and fullness which provide valid markers of the intensity and rate of change of satiety.11 Satiety is also influenced by physiological actions of consumed food especially effects in the stomach12 and the hormones released from the gastrointestinal tract during the digestion and absorption of foods.13 Although there is a view that satiety is encoded in specific peptide hormones such as CCK or GLP-1, recent studies suggest that there is no single peptide or any unique peptide profile that is responsible for satiety.14 This investigation demonstrated that a similar degree of inhibition of food intake (satiety) can be induced by quite different patterns of peptide hormone release. The profiles in Figure 3 indicate that insulin, PYY, GLP-1 and CCK were selectively released by foods high in either fat or carbohydrate – thus creating different peptide patterns. However, the degree of suppression of hunger, and subsequent food intake, were similar. We can envisage that different peptide combinations – reflecting the varying physiological actions of different foods – can all deliver the same sensations and the same degree of satiety. However, this does not mean that individual peptides do not play separate and distinctive roles. For example, both GLP-1 and ghrelin are closely associated with changes in the feelings of hunger and fullness and therefore these peptides may exert a stronger effect on short-term appetite than some other peptides. There is also a considerable body of evidence that PYY is implicated in satiety and, ultimately, in the control of body weight.15–17
Foods and the Satiety Cascade
It has been noted above that satiation and satiety are quite complex states constructed from a series of overlapping and interacting factors. Having identified some of the key processes involved in the Satiety Cascade, one natural and important question is: how do different foods impact on these processes? Do some foods possess nutritional properties that allow them to have a stronger effect than others, and therefore a better profile for controlling appetite? This is a key research area and much attention has been directed towards the macronutrients and to the issue of whether fats, proteins or carbohydrates have equally potent actions. In addition, what is the effect of dietary fibre or the glycaemic index? Do solid foods and beverages have equivalent effects? In short, how are the properties of foods related to their capacity to influence satiation, satiety or both, and therefore to influence the amount and type of energy ingested? One currently significant issue is whether it is possible to technologically engineer functional foods that will help people to manage their appetites and therefore their body weight.19 In Europe, the European Food Safety Authority (EFSA) has been established, in part, to assess the strength of the evidence on which such satiety claims for foods are based. This is a huge and complex issue which cannot be adequately summarized in this short review. However, details can be found elsewhere20 and there is an agreed methodology available to examine the effects of foods on the Satiety Cascade.21 Moreover, it is safe to conclude that not all foods produce equivalent effects and it is therefore possible to construct a diet (or meal) by selecting those foods that will deliver the most potent control over satiation and satiety.22 However, it is a separate issue whether or not people can be motivated to eat such a judicious diet. In an obesogenic environment in which there is a strong promotion of purchasing and eating, many cultural forces oppose (and often undermine) an individual’s intention to gain control over their own eating behaviour. The range of the obesogenic culture means that it will be very difficult for any single food to exert an effect on the Satiety Cascade sufficient to have an effect on the whole diet and on daily energy intake – in the absence of additional behavioural control procedures.
Homeostatic and hedonic influences
Traditionally, homeostatic regulation has been associated with the classical approach of Claude Bernard on the regulation of the internal milieu developed during the nineteenth century.23 In the field of appetite, it is used to explain the quantitative changes in eating and food intake such as those that occur during the operation of the Satiety Cascade. Curt Richter subsequently used the term behavioural regulation of internal states to indicate how eating behaviour operated to maintain physiological functioning.24 Later, Jacques Le Magnen continued the tradition from the Institut Claude Bernard in Paris. Hedonic aspects of appetite are those concerned with the influence of pleasure and palatability on eating. The homeostatic system comprises a network of gastrointestinal peptides and brain neurotransmitters and integrates physiological signals of energy requirements with the motivation to eat via sensations of hunger and fullness. The hedonic system, coordinated by the brain’s reward circuitry (that is, the network subserving pleasurable aspects of eating) responds to sensory properties and thoughts about food that reflect the cognitions of liking and wanting.25 Just like in the homeostatic system, a network of brain pathways and neurotransmitters mediates hedonic aspects of eating. It is known that these pathways are also involved in the addictive response to drugs such as cocaine and amphetamine. Although the homeostatic and hedonic systems are based on distinctive neurotransmitter systems, there is considerable functional overlap between the two systems in the control of appetite.26
Liking versus wanting food
Recently, a good deal of progress has been achieved in understanding the hedonic response to food and how separate psychological components of food liking and wanting feature in overconsumption and obesity. Clearly, the hedonic response has the potential to influence satiation (meal size) and satiety.
Kent Berridge’s27 influential theory of reward (a process that allows eating food to become a reinforcing or satisfying experience) proposes that reward is not a unitary process but encompasses both an affective (emotional) pleasure component and a non-affective motivational component, termed liking and wanting, respectively (Figure 4). Liking underpins the subjective pleasure elicited by food. Food liking is generated by the release of endogenous opioids in response to food acting on localized clusters of neurons, many in the arcuate nucleus, termed hedonic ‘hot-spots’.28 To date, hedonic hot-spots have been identified in the various regions of the brain, indicating that the rewarding (pleasurable) aspects of eating are widely distributed. This confirms their significance in the control of eating. While food liking has been described as a kind of ‘pleasure gloss’ actively painted onto the sensory perception of food,29 wanting accounts for the expression of behaviour that does not normally have a verbal or explicit explanation (e.g. ‘I do not know why, I just had to have it!’). Liking and wanting have distinguishable substrates in the brain, and while hedonic aspects of food intake are typically a combination of liking and wanting, by measuring these components separately in behaviour, we can understand how this determines an individual’s tendency to overconsume and to gain weight.30
While liking and wanting as dissociable components of reward have acquired some validity from neurobiological processes encoded in the brain, the terms liking and wanting are also derived from the semantics of a shared language that describes human activities. Liking and wanting have separate meanings in English and various European languages. Indeed, the technical definitions of food liking and wanting correspond to the ordinary understanding of these terms (and their synonyms) in the context of human appetite behaviour, such as food choice and food intake. Wanting may refer to subjective states of desire, craving, or literally to feel a
lack [of] something desirable or necessary (esp. a quality or attribute)
Oxford English Dictionary, 1989
Liking is usually understood as the perceived impact of a food or its sensory properties on subjective sensations or some judgement of the pleasure it elicits.
The nature of subjective forms of liking and wanting is that they are consciously experienced and subject to interference from other thoughts and subjective states. However, not all behaviour is under the direct control of conscious events, and therefore these components can have both an explicit and an implicit dimension (i.e. one that is conscious and one that is covert). The influence of implicit processes of liking and wanting on behaviour does not require conscious experience and may be more automated and related to behaviour where conscious awareness is low. Therefore, liking and wanting can be viewed as psychological components of reward operating at implicit (subconscious, automatic) and explicit (conscious, introspective) levels.
In the natural expression of appetite that manifests itself in a particular pattern of eating, there will be a seamless interplay between homeostatic and hedonic processes. The pleasurable response to food that may stimulate consumption will be integrated into the satiation and satiety processes that inhibit eating. Although the actual intermingling of the neurotransmitter pathways for the two systems is extremely complex, it is sufficient here to simply recognize that the expressed pattern of eating is dependent upon influences from both systems (Figure 4).
What is the driver of food intake?
Over 50 years ago, animal research gave rise to the idea that the drive to eat originated in an excitatory centre of the hypothalamus. This idea generated considerable research activity but made the source of the drive to eat rather inaccessible – particularly for dealing with human food intake.
Two notions seem to dominate the field; the first of these is the adipocentric concept of appetite control, i.e. the view that adipose tissue is the main driver of food intake, with day-to-day food intake controlled in the interests of regulating body weight (specifically, adipose tissue). During the 1950s, three basic ideas monopolized approaches to ‘body weight regulation’; these were the glucostatic,31 aminostatic32 and lipostatic hypotheses.33 These simple ideas exerted a mild but pervasive influence on thinking about a complex problem. The discovery of leptin in 1994 by Zhang et al.34 seemed to provide conclusive proof of the authenticity of the lipostatic hypothesis (which was based on interpretations of the classic rate studies of Kennedy),33 and leptin was construed as ‘the lipostatic signal’ that was an essential component required in a negative feedback process for the regulation of adipose tissue. This idea has been incorporated into models of appetite control in which leptin is depicted as the major signal (the missing link) that informs the brain about the state of the body’s energy stores.35,36 In turn, a forceful interpretation of this view has positioned adipose tissue at the centre of appetite control. Indeed, it has been stated that
There is compelling evidence that total body fat is regulated. . .when it is decreased reflexes restore it to normal. . .when it is increased reflexes. . .elicit weight loss.
Woods SC, Ramsay DS. Food intake, metabolism and homeostasis. Physiol Behav 201137
This view has been incorporated into general thinking about the control of appetite and appears to have been widely accepted. In addition, leptin is understood to play a key role in the control of appetite by adipose tissue. Although it is beyond doubt that leptin exerts a critical influence in many biochemical pathways concerning physiological regulation,38,39 it has been argued that the role of leptin in the aetiology of obesity is confined to very rare situations in which there is an absence of a leptin signal.40 Others have also argued that the role of leptin signalling is mainly involved in the maintenance of adequate energy stores for survival during periods of energy deficit.41 This is why leptin may be critical in the resistance to weight loss with dieting. More importantly for this review, there is little evidence of a role for leptin in day-to-day appetite control. In addition, the impact of adipose tissue itself has not been shown to exert an influence over the parameters of hunger and meal size, which are key elements in day-to-day control of appetite.
The second issue that appears to influence thinking is the notion of ‘energy homeostasis’. This idea has been proposed to account for the accuracy in which energy balance is maintained over time in normal individuals. A recent commentary has argued that
for a healthy adult weighing 75 kg typically consuming approximately one million kcal each year, then a mismatch of just 1% (expending 27 kcal per day fewer than consumed) will yield a body fat increase of 1.1 kg after 1 year.
Schwartz MW. An inconvenient truth about obesity. Mol Metab 201242
However, given the worldwide epidemic of obesity, and the apparent ease with which many human beings appear to gain weight, it seems implausible that some privileged physiological mechanism is regulating body weight with exquisite precision. If such a mechanism existed it would surely operate to correct weight gain once it began to occur. The compelling phenomenon of dietary-induced obesity (DIO) in rats also suggests that physiology can be overcome by a ‘weight-inducing’ nutritional environment, and that ‘energy homeostasis’ cannot prevent this. The phenomenon of DIO in rats questions the notion of an all-powerful biological regulatory system. Moreover, this experimental ‘fact’ strongly resonates with the proposal of a human ‘obesogenic environment’ that promotes weight gain in almost every technologically advanced country on the planet.2
The argument for body weight stability is not compelling. The existence of worldwide obesity suggests that body weight is not tightly regulated. An alternative view that has been discussed for decades is that regulation is asymmetrical.44 Whilst the reduction in body weight is strongly defended, physiology does not resist an increase in fat mass.45 Indeed, the physiological system appears to permit fat deposition when nutritional conditions are favourable (such as exposure to a high-energy diet). This means that the role of culture in determining food selection is critical. In many societies the prevailing ideology of consumerism encourages overconsumption. This applies not only to foods but to all varieties of material goods. The body is not well protected from the behavioural habit of overconsuming food; processes of satiety can be over-ridden to allow the development of a positive energy balance. This has been referred to as ‘passive overconsumption’4,46 and is regarded as a salient feature of the obesogenic environment.41
Updating the formula for appetite control – an energy balance approach
Over the course of 50 years, scientific thinking about the mechanisms of appetite control has changed dramatically. In the 1950s and 1960s, the hypothalamic ‘dual centre’ hypothesis was believed to provide a comprehensive account of the initiation and inhibition of food intake [e.g. Anand and Brobeck (1951)].47 Following technological advances in the identification of neurotransmitter pathways in the brain, the two-centre hypothesis was replaced by a model based on catecholaminergic and serotonergic aminergic systems.48 A recent conceptualization has proposed a theory of appetite control based on an interaction between adipose tissue (and prominent adipokines) and peripheral episodic signals from intestinal peptides.35 This two-component approach apparently summarizes current thinking. However, the history of the physiology of appetite control illustrates that any model can be improved by new findings and that some models have to be completely replaced following the advent of new knowledge. Therefore, the current conceptualisations should not be regarded as permanent fixtures; they are transient representations of the current state of knowledge.
Moreover, the current model of appetite control (based on the interaction of adipose tissue and gastrointestinal peptides) has been compiled on the basis of evidence from studies directly on the brain (of rats and mice), in vitro molecular studies on adipose tissue and experiments on peripheral hormones, such as insulin and other gastrointestinal peptides. Not since the work carried out by Edholm et al.49 and Edholm50 and Mayer et al.51 in the 1950s has thinking about appetite control taken account of evidence in the field of human energy balance research. Therefore, it is worth considering whether or not any light can be shed on the expression of human appetite from an energy balance approach.
A recent approach to the study of appetite control and energy balance has used a multilevel experimental platform in obese humans;52 relationships between body composition, resting metabolism, substrate oxidation, gastrointestinal peptides, sensations of appetite and objective measures of daily energy intake and meal size, have been examined. Such an explicit multilevel approach has not previously been undertaken. An important feature of the approach is that all variables have been objectively measured and quantified. This is particularly important in the case of daily energy intake for which self-report or -recall do not provide data of sufficient accuracy to be used in assessments of the energy balance budget.53
In several cohorts of obese people (men and women) the relationship between meal sizes, daily energy intakes and aspects of body composition (FM and FFM) have been measured simultaneously in the same individuals at different time intervals several months apart.54 Contrary to what many would have expected, a positive association was observed between FFM and daily EI, and also with meal size. In other words, the greater the amount of FFM a person has, the greater the daily energy consumed and the larger the individual meal size (in a self-determined objectively measured eating opportunity). There was no relationship with body mass index (BMI) or with the amount of adipose tissue (FM) suggesting that, in a free-running situation (with participants not subject to coercive weight loss or dietary restriction), in obese people, FM does not exert control over the amount of food selected in a meal, or consumed over a whole day. This outcome is clearly not consistent with an adipocentric view of appetite control. Moreover, the relationships were independent of gender. This means that gender does not explain the association of FFM with EI. On the contrary, FFM can explain the gender effect; men (in general) eat more than women because they have greater amounts of FFM. This observation also means that in obese people their large stores of adipose tissue do not help – but rather hinder – the control of appetite.
This association between FFM and eating behaviour has implications for an energy balance approach to appetite control, and for the relationship between EE and EI as described by Edholm et al.49 and Edholm.50 It is well established that FFM is the primary determinant of RMR, and that RMR is the largest component of total daily energy expenditure.55 From a homeostatic standpoint, an ongoing and recurring drive to eat arising from the physiological demand for energy (e.g. RMR) appears logical, as this energy demand would remain relatively stable between days and would ensure the maintenance and execution of key biological and behavioural processes. Consequently, it might be predicted that RMR, the major component of daily EE (60–70%) could be associated with the quantitative aspect of eating behaviour and with daily EI. When this was examined,56 it was demonstrated that RMR was a significant determinant of the size of a self-determined meal, and of daily energy consumed (when measured objectively and quantified – Figure 5). In addition, RMR was associated with the intensity of hunger objectively rated on hand-held electronic data capture instruments.57
Consequently, these findings – which are broadly consistent with the early predictions of Edholm et al.49 and Edholm50 – have demonstrated an association between the major components of daily EE and daily EI. In other words, they demonstrate that appetite control could be a function of energy balance. Importantly, the major findings have been replicated in completely independent large data sets that included participants from different ethnic groups showing a huge range of energy intakes,58 and from participants of variable BMIs allowed to freely select their own diet under meticulously controlled semi-free living conditions (Stubbs, Whybrow, and Horgan, Rowett Research Institute, University of Aberdeen, 2014, personal communication). These confirmatory reports suggest that the associations are robust and are not restricted to a particular group of people measured in a specific geographical location.
Considering the strength of the associations, these findings have implications for the role of FFM and RMR in appetite control. They suggest that the conventional adipocentric model should be revised to allow for an influence of FFM – in addition to FM. The adipocentric feature of the conventional model would be lessened. Our findings do not imply that FM does not play a role in appetite control. Our interpretation is that, under normal weight conditions, FM has an inhibitory influence on food intake but the strength of this tonic inhibition is moderated by insulin and leptin sensitivity.9 As people overconsume (because of cultural obesogenic influences), FM increases and the consequential increase in leptin and insulin resistance weakens the inhibitory influence of FM on appetite. This amounts to a ‘dis-inhibition’, so that accumulating FM fails to suppress food intake and permits more eating (overconsumption). Indeed, there is good evidence that low insulin sensitivity reduces post-prandial satiety and weakens meal-to-meal appetite control.59 In addition, clear positive associations of FFM and EI, and negative associations of FM and EI, have been demonstrated – but overlooked – in a comprehensive analysis carried out by Lissner et al.60 more than 25 years ago. Therefore, on the basis of these recent findings a conjoint influence of FFM and FM on appetite control has been proposed.7 This is set out in Figure 6. What are the implications of this formulation for the relationship between exercise and appetite control?
Physical activity, appetite and obesity – what are the facts?
In the surveys so far carried out the time spent sitting has varied from 83/4 to 103/4 h/day . . . It looks as though man should be regarded now, if not in the past, as a predominantly sedentary rather than an upright animal.
Edholm OG, Fletcher JG, Widdowson EM, McCance RA. The energy expenditure and food intake of individual men. Br J Nutr 195549
For a number of reasons, both theoretical and practical, it is important to clarify the effect of physical activity (and exercise) on energy intake (appetite control). The issue can be approached by addressing two specific questions formulated in classical investigations in the field. First, Edholm50 and Edholm et al.49 sought to establish a fundamental relationship between EE and EI. Edholm50 stated that the desire to find out more about the mechanisms which relate intake to expenditure is in fact what regulates appetite.
In time-consuming studies on army cadets, measures of energy expended in daily activities and energy consumed in meals and snacks showed no meaningful association within a single day. However, over a 2-week period, there was a clear association between energy intake and expenditure. Furthermore, Edholm50 also argued that the differences between the intakes of food must originate in the differences in energy expenditure.
Strangely this view was ignored and the general approach to the issue was largely abandoned. Indeed, the strategy of integrative physiology was replaced with molecular biochemistry as a form of enquiry into biology and behaviour. However, just because the views of Edholm et al.49 have been overlooked does not mean that they were wrong.
A second approach arose from the work of Mayer et al.51 and especially from his painstaking studies on jute mill workers in Bengal. In this Herculean study, a comparison was made between the physical exertion and effort required by particular forms of work and the calculated dietary intake of individual workers. Jobs ranged from heavy-duty tasks such as lifting and sorting to clerical duties and administrative desk jobs requiring little physical effort. It was assumed the energy expended was closely related to the physical effort of the daily work. In a classic figure from the published study, an inverted U-shaped function described the relationship between EE and EI. Interestingly, the right-hand portion of the curve showed an approximately linear relationship between EI and EE but only above a certain level of energy expenditure. This was consistent with the approach demonstrated by Edholm50 and Edholm et al.49 Indeed, Mayer et al.51 proposed that
. . .the regulation of food intake functions with such flexibility that an increase in energy output due to exercise is automatically followed by an equivalent increase in caloric intake. . .
Mayer J, Roy P, Mitra K. Relation between caloric intake, body weight, and physical work: studies in an industrial male population in West Bengal. Am J Clin Nutr 195651
However, Mayer et al.51 also demonstrated that at very low levels of EE – and in work that could be regarded as sedentary – the association between EE and EI was lost and dietary intake increased disproportionately in relation to the energy expended. In this ‘sedentary zone’ restraint over appetite appeared to be lost. This observation appears to have considerable implications for our current sedentary lifestyles and levels of obesity.
Interestingly, this picture has been translated into formal terms by Henry Taylor who related the homeostatic control of appetite to the physical activity performed.
. . .the late Henry Taylor favoured a model [in which] energy intake was in exact homeostasis with energy expenditure under conditions of high expenditure. . .bodily signals go awry in sedentary lifestyles; when. . .the body will not recognise it is being overfed.
Jacobs DR. Fast food and sedentary lifestyle: a combination that leads to obesity. Am J Clin Nutr 2006, p. 18961
Common perspectives about physical activity
Currently there appears to be considerable ambiguity concerning the usefulness of physical activity for weight loss, and this questions its value for dealing with the high prevalence of obesity. On the one hand, the public health authorities advise citizens to increase levels of physical activity and to decrease time being sedentary (in addition to restraining dietary intake). On the other hand, in recent years, the media has promulgated messages such as ‘Exercise will not make you thin,’62 ‘Exercise makes you fat’63 and ‘Why running makes you fat’.64 Given these strident claims it would be surprising if citizens were not confused. Since many people are pleased to read messages about the futility of exercise to reinforce a preference to avoid physical activity, the implication of this publicity is serious. Importantly, these messages portrayed in the popular media are false. There is clear evidence from large-scale controlled trials that physical activity carried out over long periods of time produces a dose-dependent reduction in body weight.65,66 The effect is clearly present whether exercise is measured in minutes of activity per week or in kilocalories of energy expended. On average, the more exercise carried out, the greater the weight loss. This evidence is supported by results from a series of reviews,67–69 including a Cochrane Review (the gold standard in assessing evidence) by Shaw et al.,70 which concluded that
exercise has a positive effect on body weight and cardiovascular risk factors in people with overweight and obesity, particularly when combined with a diet.
Shaw K, Gennat H, O’Rourke P, Del Mar C. Exercise for overweight or obesity. Cochrane Database Syst Rev 200670
A further systematic review concluded that
. . .for people starting an exercise programme, this leads to a negative energy balance and a remarkably consistent loss of body fat in relation to the net cost of exercise training.
Elder S, Roberts S. The effects of exercise on food intake and body fatness: a summary of published studies. Nutr Rev 200771
Furthermore, it is well established that the health benefits of regular exercise are independent of any changes in body weight.72
While such findings indicate that exercise can have a positive effect on body weight (if sufficient energy is expended) these reviews do not directly address the original question of Mayer et al.51 regarding whether increases in EE automatically result in compensatory increases in EI. The fact that weight loss is seen with regular aerobic exercise suggests that any ‘energy-saving’ mechanisms do not completely nullify the effects of exercise. Nevertheless, in many cases the degree of weight lost is somewhat less than that theoretically predicted on the basis of the measured EE and its presumed relationship to tissue lost.73,74 However, note that the prediction of weight loss based on such methods has recently been revised.8,43 Moreover, the changes in weight alone do not reveal the mechanisms involved, nor do they identify the physiological processes that produce the changes in body weight. The answer to these issues requires studies on appetite control to be carried out simultaneously with measures of energy expenditure. For many years the study of appetite control and the study of physical activity have been conducted quite independently – in separate specialized appetite or exercise laboratories – and the changes in EI and EE observed were interpreted in isolation. More recently, appetite research has been embraced within an energy balance framework.75–77
Studies on the impact of exercise on energy intake
It should be recognized that studies on the effect of exercise (energy expenditure) on energy intake (appetite) are the converse of studies on the effect of manipulating energy intake on energy expenditure. Both strategies intervene in an actively regulated physiological system. Considering the impact of a coercive increase in food consumed, a landmark study by Levine et al.78 has demonstrated that a mandatory ingestion of 800 kcal/day above energy requirements over a 2-month period led to an increase in body weight. However, the system tended to oppose this action through an increase in behavioural activity – called non-exercise activity thermogenesis (NEAT). However, the most striking feature of the study was the wide range of individual responses; some partcipants showed a large increase in NEAT and therefore gained little weight, whereas the opposite was true for others. This outcome calls to mind the results of the Quebec feeding study on monozygotic twins.79 Although members of the twin pairs were equally overfed, the variation in weight gained between pairs of twins was large (whereas the variation within any pair of twins was small). Again individual biological variability was pronounced.
What is the case for exercise? Many studies that have assessed EI during the manipulation of exercise have been acute in nature, i.e. often single-dose, single-day experiments (for a review, see Hopkins et al.80 or Donnelly et al.81). The clear outcome is that exercise has little effect on EI within a single day.82 However, as the exercise is continued over several days the system begins to respond and a small compensatory rise in EI has been observed in both men and women.76,83 Comparisons between participants undergoing high-, medium- and low-volume sessions of exercise indicated a graded and proportional (but partial) compensatory increase in EI which accounted for approximately 30% of the EE.76 However, there was a large range of individual variability. This variation became clear when daily exercise sessions were continued for 16 days with participants showing between 0% and 60% compensation in EI for the exercise EE.83 As anticipated, this variable response was reflected in small changes in body weight.
For medium-term studies, in which mandatory exercise sessions were performed daily for 12 weeks in overweight and obese individuals,73 an average weight loss of approximately 3.3 kg was recorded (the average reduction in body fat was also 3.3 kg) but with weight change varying between –14.7 kg and +1.7 kg. This outcome is quite remarkable because the weight gain of some participants was achieved despite the performance of supervised and measured exercise sessions (5 days per week for 12 weeks). Therefore, even though all participants completed the exercise sessions (with total exercise-induced EE calculated at 28 000–29 000 kcal), there was large variation in the change in body composition. The variability in body weight changes following 12 weeks of supervised aerobic exercise has subsequently been replicated in a larger number of overweight and obese individuals (see Figure 7) and in several other trials of the effects of exercise on weight loss in obese people.
In many studies on the effect of exercise on body weight, the average weight change would be regarded as the most important parameter. However, as several writers, such as Dilnot and Blastland,84 have pointed out, science is often weakened by subscribing to the ‘tyrany of the average’. The most significant outcome of the study by King et al.73 – and other similar investigations74 – is the range of individual adjustments in body weight as shown in Figure 7. This type of outcome is robust and has been demonstrated in different types of participants followed over similar time periods.74 However, more significant than the change in body weight is the effect of exercise on body composition. The weight lost is almost entirely adipose tissue, where as the weight gain is reflected in lean mass (FFM), which is apparent in both men and women.85
Exercise and the appetite control system
Can the effects of exercise on body composition be explained by actions on the appetite control system? First of all, it is clear that any compensatory increase in EI, which could offset the exercise-induced EE, is not uniform. Compensation varies markedly from person to person. One reason for this is the observed effects of exercise on different components of appetite control. During medium-term studies it has been shown that exercise exerts a dual control of the expression of hunger.86,87 There is an exercise-induced increase in fasting – or early morning – hunger. However, in parallel, exercise also improves satiety by increasing the post-prandial sensitivity to ingested nutrients consumed in meals. Interestingly, the increase in post-prandial satiety – measured by the satiety quotient which is an index of the satiating capacity of the energy consumed88 – is shown by all people who perform the exercise. However, the effect of exercise on fasting hunger is quite variable. Therefore, it can be deduced that the range of the effects of exercise on overall EI is a function of the individual change in basal hunger together with the adjustment in post-prandial satiety.86 The increase in post-prandial satiety is quite consistent with the effect of prolonged exercise on the fasting levels of gastrointestinal peptides; a decrease in ghrelin but an increase in CCK, GLP-1 and PYY.
A theoretical issue is whether the action of continuous exercise can be understood in the light of recent findings concerning the physiology of appetite control as described above. First, the objectively measured responses in appetite behaviour which in turn change EI can be accounted for by the impact of continuous exercise on physiological processes. Since exercise produces adjustments in blood flow, gastrointestinal hormone response, gastric emptying, muscle cellular metabolism, adipose tissue biochemistry and brain activity, it will inevitably interfere with several of the mechanisms involved in the episodic control of appetite. Acute responses to exercise include changes in ‘appetite’ hormones such as ghrelin, GLP-1 and PYY as noted above89 as well as variable changes in substrate oxidation in muscle and liver which may related to the post-exercise change in hunger and food intake.90
Second, as exercise is repeated over months effects on body composition are observed. These, normally, take the form of a decrease in FM with maintenance of, or an increase in, lean tissue (FFM).91 These more gradual changes will bring about adjustments in the tonic control of appetite. Of importance will be a change in RMR because of changes in FFM and FM – now shown to be a determinant of meal size and daily EI56 (as depicted in Figure 5), and changes in insulin sensitivity (arising indirectly from a reduction in adipose tissue), which influences the accuracy of post-prandial satiety.59
Consequently, it is possible to formulate an account of the way in which exercise can influence body weight. Acute effects of exercise on appetite will be mediated by episodic ‘satiety’ signals (arising from the act of eating, changes in substrate oxidation (during or immediately after exercise), gastric emptying or other stomach events and skeletal muscle activity (postulated to alter brain dopamine and other transmitters). The effect of enduring exercise will be mediated via changes in body composition in addition to the short-term changes noted above. Indeed, the roles of FFM and FM in appetite control seem crucial to an understanding of the action of exercise. Exercise will usually increase FFM and decrease FM. An increase in FFM will increase the demand for energy (to meet increased energy requirements) and this will involve an increase in basal hunger. A decrease in FM will lead to greater post-prandial inhibitory control of appetite (satiety) partly by an increase in insulin and leptin sensitivity. Therefore, enduring exercise will lead to an increased sensitivity of appetite control mechanisms. This means that EI will be better matched to EE. However, one consequence of this is that, overall, EI may be increased.
This formulation takes into account the initial concept of Mayer et al.51 and the proposition of Henry Taylor (as quoted by Jacobs61). As people commence and maintain physical activity there is an increase in FFM and an improvement in post-prandial signalling (appetite control). This means that appetite is better regulated. However, in sedentary states (80–90% of the population) when EE is low, EI is not down-regulated to match the lower EE. These sedentary individuals also show a high level of opportunistic eating (high scores on Three Factor Eating Questionnaire disinhibition) and a susceptibility to overconsumption and therefore to weight gain. This means that EE and EI are not independent; they interact – physical activity influences appetite control and changes in dietary intake can influence behavioural activity.
A summary of the impact of exercise on the control of appetite is set out in Figure 8. This formulation indicates how the cumulative effect of exercise on body composition (FM and FFM) with implications for hormone sensitivity, together with changes in gastrointestinal peptides responsible for satiety signalling can lead to variable modulation of the compensatory response. The effect of the particular intensity and duration of exercise on an individual person’s change in body tissues or hormone release, would lead to specific adjustments in the motivation to eat and the satiating response to foods consumed. Consequently, any compensation to prolonged exercise will depend, to a large degree, on the variability of the biological responsiveness between individuals. Therefore, the compensation can be accounted for by the action of exercise on the physiological mechanisms of appetite control. In turn, the biological variation in these mechanisms from person to person can account for the variable effect of exercise on body weight (and body composition).
Individual variability – a key feature of energy balance behaviours
One key lesson quickly learned (but seldom declared) from working within a psychobiological framework is that aspects of human appetite expression are characterized by huge individual differences. One message emerging from the genetic analysis of appetite is that there is huge allelic variation (genetic differences) operating on the processes that influence food intake and energy balance. This diversity poses a considerable problem for devising public health policies for dealing with eating behaviour change in relation to the obesity epidemic.92
The description of susceptible and resistant phenotypes (together with other research not mentioned here) has drawn attention to the wide diversity in the pattern of the human eating response in the face of an obesogenic culture. Perhaps this should not be surprising given the great variability in the nutritional patterns adopted by the human species in quite extreme ranges of climate and habitat. What are the implications of this diversity? One methodological issue concerns the use of the statistical mean – or other measures of central tendency – to describe responses to interventions or treatments. Very often the mean outcome fails to adequately reveal the true effect of the intervention or treatment (the weight loss response to enforced exercise is a good example). A truer reflection of the operation of the challenge is described by the diversity of responses that encourages a deeper examination of the internal processes responsible. In other words, this means that one unique explanation cannot account for all outcomes.
This issue draws attention, once again, to the nomothetic and idiographic approaches to scientific explanation.93 What should be the balance between seeking a common unifying principle and a regard for individual differences (quantitative and qualitative)? In light of this question, it may be an appropriate time for a paradigm shift (along the lines envisaged by Kuhn94) to focus attention on individual variability rather than on the mean value of any set of responses. In scientific research, the mean response is the statistical parameter associated with the elucidation of scientific principles. However, given cause–effect relationships and other features normally seen as the objectives of scientific inquiry, the great diversity of the human eating response suggests that we are dealing with a phenomenon for which the average is often inappropriate. This means that the traditional use of normal scientific methodology in research may be missing much that is truly important in defining human appetite. The identification of phenotypes – their behavioural expression and their underlying physiology and genetics – constitutes a partial step towards a recognition of the variability inherent in human energy balance behaviours.