Disease is a ubiquitous and powerful evolutionary force. Hosts have evolved behavioural and physiological responses to disease that are associated with increased survival. Behavioural modifications, known as ‘sickness behaviours’, frequently involve symptoms such as lethargy, somnolence and anorexia. Current research has demonstrated that the social environment is a potent modulator of these behaviours: when conflicting social opportunities arise, animals can decrease or entirely forgo experiencing sickness symptoms. Here, I review how different social contexts, such as the presence of mates, caring for offspring, competing for territories or maintaining social status, affect the expression of sickness behaviours. Exploiting the circumstances that promote this behavioural plasticity will provide new insights into the evolutionary ecology of social behaviours. A deeper understanding of when and how this modulation takes place may lead to better tools to treat symptoms of infection and be relevant for the development of more efficient disease control programmes.
‘Yo soy yo y mi circunstancia’ (‘I am myself and my circumstance’, , p. 43)
Disease is a ubiquitous and powerful evolutionary force. As a result, hosts have evolved behavioural and physiological responses that increase survival under infection . Behavioural modifications, known as ‘sickness behaviours’, frequently involve symptoms, such as lethargy, somnolence and anorexia. While historically disregarded by physicians as mere side effects of the disease, sickness behaviours are increasingly viewed as an adaptive response [3,4]. A shared feature of these behavioural responses is an overall reduction in activity, which is thought to preserve metabolic resources that can then be allocated to fighting the infection [3,5] and, in the case of anorexia, to reduce nutrients that are essential for bacterial growth [5–7].
Life-history theory predicts that when circumstances are such that survival and reproduction are in conflict, the amount of investment in sickness behaviours should shift (figure 1). The question is thus: ‘when should animals experience symptoms of sickness and when should they suppress them?’ In this review, I focus on research demonstrating that a diversity of social contexts can act as potent modulators of sickness behaviours. Exploring these reports brings us closer to understanding what conditions lead infected animals to alter investment in sickness behaviours. The study of this behavioural plasticity can give us new perspectives on social behaviour. While appropriate social interactions are relevant for fitness, they also create opportunities for disease transmission. I argue that furthering our understanding of when and how the modulation of sickness behaviours occurs might be relevant for improving the health and management of animals in captivity and the control of disease spread in captive and wild populations. Finally, I point out clinical implications derived from understanding the neuroendocrine control of this phenomenon.
2. Behavioural effects of infection
Sickness behaviour is defined as a generalized reduction in the occurrence of an array of behaviours in response to an infection. The changes include reduced food and water intake, reduced activity, reduced engagement in social activities, decreased exploratory behaviour, inability to experience pleasure, decreased libido and increased somnolence (summarized in ). Sickness behaviours are widespread in the animal kingdom, occurring in invertebrates  and a variety of vertebrates, including amphibians , reptiles , birds  and mammals, notably humans .
Importantly, sickness behaviours are not caused by the infectious agent itself, but by the host organism as it responds to this agent via central and peripheral release of cytokines, which include interleukin-1 (IL-1), interleukin-6 (IL-6) and tumour necrosis factor alfa (TNFα)  (but see recent report by Chiu et al.  on effects of bacteria on pain receptors). Accordingly, administration of IL-1β and TNFα induces sickness behaviours in rodents (reviewed in ). Another widely used method to mimic the symptoms of an infection that permits focusing on host-mediated (versus pathogen-mediated) effects on behaviour is via administration of non-pathogenic antigens. One example is lipopolysaccharide (LPS), a component of Gram-negative bacteria cell walls, which induces sickness behaviours by activation of the immune response . The availability of these tools (both cytokines and non-pathogenic antigens) makes the study of sickness behaviour very tractable and easy to implement.
3. Sickness behaviours, plasticity and motivation
Historically, physicians have regarded sickness behaviours as an undesirable side effect of disease (as described in ). Hart  proposed an alternative view, suggesting that these behaviours consist of a highly organized strategy to aid in fighting the infection, by shifting energy from non-essential activities into the immune system. From this framework, emerges the possibility of a trade-off: the body has limited resources (which could be in the form of energy, nutrients or time) and in order to fight an infection these need to be allocated towards sickness behaviours at the expense of other behaviours. The idea of a trade-off relies on the immune response being costly, and there are currently several lines of evidence for energetic costs of immunity [18–20]. These costs can be converted into reduced growth , reproduction  and survival . When animals are infected, there should therefore be a region of optimal investment in sickness behaviours that maximizes lifetime reproductive success. The ability of individuals to modify their behaviour according to different environmental conditions (i.e. to exhibit behavioural plasticity) should be of great importance in balancing the costs and benefits of this trade-off.
Early studies by Miller  suggested that sickness behaviours could be considered a motivational state. In Miller's initial experiment, he demonstrated that when undergoing an endotoxin challenge, rats had higher motivation to rest than to receive rewards or avoid electrical stimulation. This observation indicates that, when sick, animals reorganize their priorities and are able to adjust behaviours in a way that benefits their recovery from infection. This reorganization may become more complex when infected animals are simultaneously faced with numerous competing factors, which probably happens in their natural environment. Since the initial motivational approach, several studies have shown that a variety of factors can affect the extent to which sickness behaviours are expressed, including abiotic factors such as season (e.g. [25–27]) and biotic factors such as the animal's sex [28–30].
4. Social context: a potent modulator of sickness behaviours
An animal's social context encompasses many factors that have the potential to interact to impact the motivation to engage in sickness behaviours. Social behaviour is defined here as behaviours expressed towards conspecifics that benefit one or more individuals in the group and social context as the social setting (including social structure of the group and social roles within it) in which these behaviours take place (sensu ; also see ). Because social context is a main determinant of the costs and benefits associated with investment into survival versus reproduction, it should greatly affect the amount of investment in sickness behaviours (figure 1). Considering the importance for fitness of fine-tuning behavioural expression to changes in social environment , the terminal investment hypothesis  would suggest that these types of behavioural adjustments should be even more critical during an infection when animals experience deteriorating physical conditions. In the following sections, I briefly review the most studied social contexts that can affect the expression of sickness behaviours (table 1).
Mating behaviour has obvious implications for fitness when it leads to production of offspring. In the context of mating, males and females should generally have different motivations to modulate their investment in sickness behaviours. While for both sexes, investment in sickness behaviours may lead to increased lifetime fitness by increasing the likelihood of survival to the next season, males that suppress symptoms of infection when presented with the chance to mate will probably gain an immediate fitness advantage, especially if mating opportunities are limited. By contrast, females that decrease mating behaviour when ill are reducing the risk of spontaneous abortion of the fetus during infection , thereby minimizing fitness losses. One example of the effect of social context on sickness behaviour is the study carried out by Yirmiya et al.  demonstrating that, when presented with a mate, male rats are less sensitive to the effects of IL-1β on mating behaviour than female rats. There is also indication that immune challenged male zebra finches (Taeniopygia guttata) are affected by the presence of a potent sexual stimulus. When presented with a novel female, animals suffering from a simulated acute infection were able to not only behave similarly to control-injected birds, but also to activate their reproductive axis and to court females to the same extent as the controls . However, experiments in male mice show an almost opposite effect, with an increase in depressive-like behaviours when presented with a receptive female, as compared to control males kept in isolation, as well as elimination of mating behaviour, as compared to control-injected mice . More studies focusing on species with different reproductive strategies are necessary to fully understand the extent and direction of behavioural modulation. The challenge will be predicting how different diseases alter the trade-off between current and future reproduction and how the social environment impacts this trade-off. One intriguing comparison would be of sickness behaviours of closely related species of semelparous (species that breed only once in their lifetime) and iteraparous (species able to breed more than once in their lifetime) animals. Given such a concentrated effort in reproduction, do semelparous species even display sickness behaviours during the breeding season? One informative type of study contrasts the same species living under different environmental conditions that modify the costs and benefits of foregoing mating under infection. For example, song sparrow (Melospiza melodia) populations at low latitudes have a longer breeding season than their higher latitude counterparts and, thus, animals of low-latitude populations should be able to invest more in recovering from infection during this season. Indeed, birds from a population at lower latitude had more intense and longer lasting sickness behaviours (assessed by activity obtained via radiotelemetry) after an LPS injection than those from the higher latitude population . Since these differences could be owing to other factors, such as immediate environmental differences (e.g. temperature), Adelman et al.  also tested animals from the two populations in a common environment under controlled laboratory conditions. They found that the behavioural differences between the populations are maintained in captivity, which points to additional factors in explaining this variation (such as genetic or maternal effects, for example).
(b) Parental care
In many taxa, postpartum parental care is essential to survival of offspring. A number of experiments suggest that the drive for parental care partially trumps infection symptoms. Aubert et al.  demonstrated that in conditions which place the survival of the offspring at risk, maternal care overcomes sickness behaviour. Specifically, when female mice injected with LPS and their litters where housed at different room temperatures, nest building in LPS-injected dams was reduced at room temperature (20°C), but not at critically low temperatures (6°C). It is a common behaviour in rodents for lactating females to display aggression towards intruders, given the likelihood of infanticide. Weil et al.  demonstrated that in mice, maternal aggression towards a virgin male intruder was not changed by an LPS injection at a dose sufficient to induce classical components of sickness behaviour (such as reduced food intake). Both of these experiments show that the expression of sickness behaviour is reduced in situations where maternal care is more critical for pup survival. To optimize fitness benefits, female investment in sickness behaviours should be higher before mating, since disease can increase chances of spontaneous abortion or damage to offspring (reviewed in [47,50]), while during parental care, sickness behaviours should be decreased if this allows for greater expression of care essential for offspring survival. This is especially true for mammals, in which offspring are more dependent on post-natal maternal care (through lactation). Animals that maintain food supplies (e.g. hoarders) can engage in sickness behaviours while still investing in future needs. Siberian hamsters subjected to an LPS injection reduce food intake but maintain hoarding behaviour . Since male, but not female, Siberian hamsters showed a trend towards decreasing hoarding, the authors suggested that females might behave this way to alleviate the energetic costs of lactation in the future. Exposure to endotoxin seems thus more effective in suppressing the immediate response to food than the anticipatory response to future needs.
There is evidence that the amount of investment in parental care during disease also depends on the potential fitness gain of the current reproductive bout. In a study of house sparrows, the likelihood of a female abandoning the nest after an LPS injection diminished with brood size . In non-mammalian species with bi-parental care (many birds, for example), sickness behaviours of both sexes should be affected to the same extent since both parents contribute the same type of resources and are equally important to ensure offspring survival. Of course, additional factors might interact to affect modulation of sickness in this scenario, as for example, the opportunity for extra pair copulations, which might bias one of the parents to being more prone to obscuring symptoms of sickness. Experiments specifically addressing how different forms of care contribute to modulation of sickness behaviours have yet to be carried out, even though there is the suggestion that sickness behaviours are differentially affected in species with different types of parental care. For example, while IL-1β administration to females of a uniparental (female-only) rodent species (Mus musculus) inhibited sexual behaviour and eliminated the expected preference to mate with intact versus castrated males , LPS injection of females of a biparental rodent species (Microtus ochrogaster) enhanced partner preference of familiar versus unfamiliar males and facilitated pair bonding . These differences might suggest that while females of species with biparental care can afford to invest in reproduction while sick, given the paternal contribution to care, the opposite behaviour might be adaptive for females that receive no help from males in caring for offspring. Indeed, male house sparrows show increased paternal effort (feeding rate) when their mates reduce feeding effort owing to an LPS injection . By exposing closely related species to a simulated infection, we may reveal the conditions that promoted differences in behavioural strategies. Social modulation of sickness behaviours can thus provide a paradigm for studying the motivations underlying social behaviours with implications for mate choice and sexual selection theory.
(c) Early social environment
Sickness behaviours are affected on both ends of the parental care dyad: from the parental side (as described above) and from the offspring side. When guinea pig pups are isolated and injected with LPS and later reunited with the mother, only the behaviour of male pups is ameliorated . This study once again emphasizes sex differences in social modulation of sickness behaviours. Early social environment can also influence sickness responses later in life. In a study by Tuchscherer et al. , domestic piglets were exposed on a daily basis to 2 h of social isolation from ages 3 to 11 days. When their response to an LPS injection 45 days after isolation was measured, the animals subjected to isolation demonstrated a significantly higher vomiting response then control animals. A similar effect was found for mouse pups separated from their dams . More recently, Avitsur et al.  showed that early age separation of mouse pups from the dams altered the proinflammatory response to endotoxin administration later in life and that these alterations seem to be sex-specific. While it might be challenging to account for in field studies, in the laboratory it may be relevant to consider early rearing environment when examining adult immune responses.
(d) Agonistic interactions: territorial intrusions and social status
Animals that are able to hold high-quality territories can have fitness advantages in several ways, for example, by having access to better food sources, better nest sites and better mates. In fact, only territorial males are able to mate in many passerine species. Song sparrows (Melospiza melodia morphna) are territorial all year round. Nevertheless, the response of male song sparrows to an LPS injection is dependent on season. Injected males reduce territorial aggression towards conspecifics during the non-breeding season, but show low responsiveness to the same dose of LPS in the early breeding phase, when territorial defence dictates reproductive success . A similar effect occurs in rhesus monkeys (Macaca mulatta), where the effects of somnolence induced by administration of IL-1 are abolished by the presence of an intruder in the enclosure . Interestingly, in contrast to the social withdrawal that is typically observed in many rodents as a result of an LPS injection [14,16], rhesus monkeys display greater affiliation towards conspecifics . This illustration of species-specific differences of selected aspects of sickness behaviours should encourage comparative studies that can help us understand how different aspects of sickness behaviours are regulated. One complicating factor that should be emphasized is that even for what may initially seem similar manipulations, like group versus isolation housing, very different demands are imposed onto the individual depending on its social structure. For example, while male rats are able to cohabit at high densities without great manifestations of aggression, group-housing can be extremely stressful for certain species of mice, in which hierarchies need to be established and maintained often through aggressive interactions.
Social rank has been shown to affect physiology and health in several taxa, including primates , but its link to modulation of sickness behaviour is less well understood. However, studies of male mice indicate that social position in a hierarchy differentially affects the expression of sickness behaviours. Dominant male mice injected with LPS showed reductions in activity and aggression, whereas submissive males exhibited increases in both defensive and social exploratory behaviours . Hence, it seems that higher social ranking is affording the dominant males the possibility to focus on recovering from an infection, whereas submissive males still need to display defensive behaviours. In zebra finches, group-housing (versus housing in isolation) is associated with reduced expression of sickness behaviour without significant alteration of the inflammatory response as quantified by plasma IL-6 , which could be an indication that in certain social contexts animals have some motivation to conceal their sickness, for example, to maintain social status. In male rats, group-housing also attenuates LPS-induced sickness behaviours, whereas in females, group-housing exacerbates these behaviours relative to LPS-injected animals housed in isolation . In an experiment using house finches (Haemorhous mexicanus), infection with the bacterium Mycoplasma gallisepticum caused individuals to become more submissive (i.e. less aggressive). The alteration in behaviour of the sick bird in turn caused healthy conspecifics to increase their proximity to these individuals, as they were found more frequently feeding near them . Changes in behaviour of both the infected and non-infected individuals could therefore have important impacts on disease transmission rates. Indeed, it was recently shown that in wild deer mice (Peromyscus maniculatus) behavioural phenotype is associated with infection status : bold mice had three times the likelihood of being infected with Sin Nombre Virus than did shy mice. While the authors cannot determine with certainty whether the infection is a cause or a consequence of the behaviours, this study alerts for the importance of behavioural heterogeneity for disease transmission.
5. Costs to suppressing sickness behaviours
The ability to modulate sickness symptoms according to the social context could prove adaptive in the sense that it might allow the animals to keep their social position in the group, preserve mating opportunities and increase the survival of offspring. But on the other hand, not giving the body the opportunity to fight the infection could have damaging effects on health. While several studies have focused on testing whether mounting an immune response is costly, fewer studies exist on the consequences of supressing the immune response, specifically concerning the behavioural part of this response. One classic example of costs of not engaging in a behavioural response associated with infection comes from experiments with ectotherms prevented from developing a behavioural fever (lizards: , fish: ). These animals suffered from higher mortality than animals that were able to develop fever. What physiological costs may accompany the socially induced suppression of sickness behaviours? A study by Lopes et al.  suggests that zebra finches which spend more time resting also have increased immune defences, as quantified by their ability to develop a fever response, to produce acute phase proteins and to kill bacteria. Understanding the consequences of social suppression of sickness behaviours could be used, for example, to improve welfare of sick animals, and it is thus a theme that warrants further exploration.
When sickness behaviours are derived from chronic inflammation, social modulation of sickness may be beneficial. It is known that certain chronic depressive disorders and dysfunctions of the central nervous system can be associated with poorly regulated and prolonged inflammation . Also, sickness behaviours and depression share similar symptoms and, potentially, similar inflammatory pathways . A study of female mice bearing ovarian carcinoma suggested that social housing was able to reduce depressive-like symptoms , as quantified by a sucrose intake test. In an experimental model of stroke in rats, social housing (males housed with ovariectomized females) reduced ischaemic damage and mortality through a mechanism that appears to be mediated by changes in the inflammatory response . While this study did not focus on behavioural alterations, it illustrates how for certain diseases social housing might be beneficial. Actions in which increased human contact can be provided for patients with these types of conditions might help ameliorate some of their symptoms.
6. Mediators of social modulation of sickness behaviours
The behavioural expression of disease should ultimately result from a crosstalk between the immune, endocrine and nervous systems. All of these systems are susceptible to being impacted by changes in the social environment. While an in-depth discussion of mechanisms is outside the scope of this review, I pinpoint some potential modulators of sickness behaviours in the context of social interactions. Obvious candidates from the immune system are pro- and anti-inflammatory cytokines. Cytokines can directly impact behaviour [29,63] and a number of experiments have demonstrated an ability of the social environment to alter these [35,36,62]. A recent review by Hennessy et al.  explores the potential for cytokines to have social functions in non-infected individuals. Regarding the endocrine system, socially modulated hormones with immunomodulatory actions are likely candidates. The most studied mediators have been testosterone and glucocorticoids. While many studies indicate a suppressive effect of testosterone on the immune system (e.g. [18,65,66] and recently ) and even directly on sickness behaviour [68–71], this is not always the case [72–75]. The story is also complex with glucocorticoids. While historically seen as immunosuppressive hormones , different lines of evidence indicate that stress can actually potentiate the immune response . Nonetheless, removal of the adrenal glands leads to increased sickness behaviours [77–80], suggesting a role for glucocorticoids in this mediation. Neural transmitters that are relevant for the regulation of social behaviours and also seem to be involved in modulating inflammatory responses, include acetylcholine, serotonin, dopamine and nitric oxide. For a comprehensive review of the bidirectional interaction between the brain and immune system, see .
Different modulators might be more relevant depending on the social context being studied. For example, in conditions that are perceived as stressors, such as in the case of isolation housing, maternal separation or territorial intrusion, the main mediator might be a glucocorticoid. By contrast, sex differences in sickness behaviours might be linked more strongly to sex hormones. Thus, to unravel the underlying mediators of the social modulation of the sickness response, it is critical to take careful consideration of the differential load that different social contexts pose on the individual.
7. Implications, directions for future research and conclusion
The study of the modulation of sickness behaviours lies at the intersection of several fields (behavioural ecology, immunology, neuroendocrinology, psychology, evolution, animal welfare, veterinary and human medicine), providing a new platform to study behaviour from diverse angles. Experiments using immune challenges to examine the flexibility of the immune response under different circumstances are contributing to advancing the understanding of life-history evolution , as well as untangling the relative contributions of non-genetic (including, for example, ecological as well as maternal effects) and genetic factors towards the observed differences in behaviour during infection [49,82]. For example, recent studies are helping to shape models of the terminal investment hypothesis . The growing interest in the ecological aspects of disease is reflected by the recent creation of a Division of Ecological Immunology and Disease Ecology within the Society for Integrative and Comparative Biology, and a Disease and Host–Parasite Ecology Section within the Ecological Society of America.
Since the findings summarized here demonstrate that expression of sickness behaviour can be modified in certain social circumstances, we should be aware of instances where we might not be able to identify sick animals. Early detection of diseased animals should help prevent disease spread and prove relevant in a world where infectious diseases cause a major burden in terms of both lives lost and economic damage . Comparative studies leading to an improved understanding of the significance of social context for different animals should facilitate our ability to predict species-specific sickness behaviours and the social cues leading to changes in these behaviours. This knowledge can then be used to determine the species and circumstances in which we can detect sickness behaviourally and the ones for which additional tools are necessary, in order to, for example, guarantee food safety and avoid spread of diseases within animal facilities.
The literature reviewed highlights the great plasticity of the sickness behaviour response in the face of social stimuli. However, we have limited understanding of the physiological consequences for the host in diminishing investment in sickness behaviours and further investigation is necessary. When making decisions regarding when to allow and when to try to suppress sickness behaviours, veterinarians, doctors and animal facility managers must weigh these costs against the potential for pain or suffering induced by the disease. If social housing can ameliorate the symptoms of non-communicable diseases with little cost to the host, then we might be easily able to increase wellness of captive animals suffering from disease.
Understanding how, when and why social modulation of sickness behaviours occurs will not only advance our knowledge of life-history evolution and the motivation for engaging in social behaviours, but also translate into better prevention strategies for infectious disease spread and improved tools to ameliorate symptoms of infection and depression. The increased connectivity of animal populations (including humans) has led to an unprecedented potential for disease pandemics. It will take concentrated and transdisciplinary efforts to understand and address these problems. In doing so, we would be wise to consider the role of social context in the modulation of sickness behaviours.
I thank the Institute of Evolutionary Biology and Environmental Studies at the University of Zurich for financial support through a post-doctoral research position.
I am grateful for comments provided by Gregory R. Goldsmith, Manuela Ferrari, Eileen A. Lacey, Samuel L. Diaz-Muñoz, Tobias Uller, Jim S. Adelman, Noah T. Ashley and George E. Bentley. Finally, I thank Barbara König and Andri Manser for discussions on figure 1.
- Received January 27, 2014.
- Accepted May 23, 2014.
- © 2014 The Author(s) Published by the Royal Society. All rights reserved.