Consumption, contact and copulation: how pathogens have shaped human psychological adaptations

Abstract

Disgust is an emotion intimately linked to pathogen avoidance. Building on prior work, we suggest disgust is an output of programmes that evolved to address three separate adaptive problems: what to eat, what to touch and with whom to have sex. We briefly discuss the architecture of these programmes, specifying their perceptual inputs and the contextual factors that enable them to generate adaptive and flexible behaviour. We propose that our sense of disgust is the result of these programmes and occurs when information-processing circuitries assess low expected values of consumption, low expected values of contact or low expected sexual values. This conception of disgust differs from prior models in that it dissects pathogen-related selection pressures into adaptive problems related to consumption and contact rather than assuming just one pathogen disgust system, and it excludes moral disgust from the domain of disgust proper. Instead, we illustrate how low expected values of consumption and contact as well as low expected sexual values can be used by our moral psychology to provide multiple causal links between disgust and morality.

This article is part of the Theo Murphy meeting issue ‘Evolution of pathogen and parasite avoidance behaviours’.

1. Introduction

Over the past few decades, scholars from across academia have shown an increased appetite for disgust. Although many explanations of disgust have been proposed, most researchers, particularly those taking an evolutionary approach, tend to agree that disgust evolved to avoid substances and behaviours that increased exposure to pathogens in ancestral environments [1–5]. Building on the work of Tybur and colleagues [5,6], we model disgust as an output of information-processing circuitries that evolved to address three separate adaptive problems: what to eat, what to touch and with whom to have sex [7].

This modest reframing of the model proposed by Tybur et al. [5] advances the field in three important ways. First, it identifies critical internal regulatory variables—expected values of consumption, expected values of contact and expected sexual values—that give rise to our sense of disgust. Second, it specifies the inputs—both internal and external—that feed into systems that compute each internal regulatory variable, as well as how these inputs are modulated by contextual factors, such as developmental stage, gender and opportunity costs, giving rise to the variability and flexibility in disgust responses that we observe across individuals and across time. Third, this reframing removes morality from the proper domain of disgust and alternatively suggests that systems regulating norm creation, adoption and enforcement use the expected values of consumption, contact and sex, thus generating several causal links between disgust and morality.

Our approach here focuses primarily on system design—that is, the specific inputs, integrators and outputs that could arguably guide decisions regarding consumption, contact and copulation and give rise to our sense of disgust (for more on emotions and internal regulatory variables, see [8,9]). Frequently, descriptions at the information-processing level of analysis violate common intuition. For instance, although we could describe the altruism represented by a mother caring for her sick child in terms of assessed probabilities of relatedness, need state and trade-offs in investment directed towards her other children, these variables are not consciously accessed. Instead, they generate the motivational force of love, care and concern. And so it goes for how we make decisions regarding what to eat, what and whom to touch, and with whom to have sex. The motivational force and feelings of extreme distastefulness versus yumminess, a longing to avoid versus be near, and sexual revulsion versus arousal can all be explained in terms of the information that natural selection brought to bear when generating those feelings and intuitions.

Here, we briefly outline the model and consider disgust's relationship with morality. Because disgust is ultimately related to pathogenic microorganisms, we begin with the selective pressures of pathogens.

(a). Ultimate selection pressures: the force of pathogens

Pathogens—bacteria, viruses and other microorganisms—though small in scale have arguably posed one of the largest selection pressures on long-lived multicellular organisms, including humans [10,11]. One reason for this is the vast difference between the rates of replication of pathogens when compared with their hosts. Whereas it takes approximately 20 years for humans to reproduce, many bacteria can reproduce every 20–30 min, amounting to approximately 350 000 bacterial generations for every one of ours [12,13]. The relative time lag in host reproduction presents pathogens with a more-or-less stable environment to which they can adapt, learning how to evade components of internal defence systems and how to capture host resources in their own struggle to survive and replicate [14,15].

Hosts in return have evolved a set of physiological and psychological defences to stem the tide of invasive microorganisms [16]. With respect to physiology, in addition to an exquisitely complex immune system, the host–pathogen arms race has led to the evolution of various design features, all of which create a moving target for invading pathogens [11]. For instance, one feature of long-lived multicellular creatures is that different genes turn on at different developmental time periods. Another feature is the presence of many different versions of each gene, or alleles. Indeed, for genes involved in the ability to recognize foreign cells entering the body there can be as many as 50–100 different alleles [17]. The horizontal variability observed in gene expression within a cell and the exact type of genes expressed across individuals (groups of cells) are examples of pathogen defence—both help to vary the array of expressed genes and gene-products, the chemical targets of microorganisms.

Natural selection also favoured mechanisms promoting vertical genetic variability—sex [11,18,19]. When compared with the clonal process of asexual reproduction, sexual reproduction generates new combinations of genes. With sex, pathogens that might have been well-adapted to a particular DNA sequence or its protein product are no longer as well-adapted after genomes recombine and produce novel—yet still functional—sequences. Sexual reproduction changes the internal biochemistry of hosts and pushes pathogens back to square one in their progress towards host domination. Importantly, the advantages of sex are obtained only if individuals recombine genomes with others who do not share the same genes. Sex with a clone is no different from asexual reproduction in that it maintains a similar environment to which pathogens can learn to adapt and exploit.

Of course, some organisms do reproduce asexually. However, this state of affairs can be maintained for a variety of reasons. For instance, asexuality could be favoured when asexual organisms have generational times that more closely match those of their resident pathogens. With lockstep reproduction, hosts can evolve countermeasures against pathogen offensive strategies much more rapidly, diffusing pathogens’ rates of adaptation. Or, asexual populations might have alternative defence mechanisms that counteract pathogens [18].

To complement the many physiological design features that promote horizontal and vertical genetic polymorphism, humans also evolved a suite of psychological and behavioural mechanisms that protect against the various threats that pathogens pose. Because the physical environment introduces avenues for pathogen transmission—via ingested organic matter and contact with substances and surfaces harbouring pathogenic microorganisms—design features that decreased the volume of pathogens the body had to fight off, without compromising too much by way of nutrition or social interaction, would have been advantageous and quickly become species-typical. Similarly, due to the direct fitness benefits of maintaining polymorphisms down the generations, psychological design features that inhibited mating with individuals who had an increased probability of sharing sets of genes, and who could thus jeopardize the production of healthy, reproductively capable offspring, would have also increased in frequency. For these reasons, we should expect to see psychological programmes that guide decisions regarding consumption, contact and sex, respectively, in a manner that adaptively stemmed the tide of infectious pathogens.

Taken together, we should expect to see mental programmes that evolved to mitigate the multiple threats pathogens pose. Following the trail initiated by evolutionary-minded scientists (e.g. [2,5,6,8]), we propose specific information-processing systems that evolved to regulate consumption, contact and sex. For each domain, we discuss the inputs, the various contextual factors that would have been critical in trading off pathogen avoidance in favour of other fitness-promoting behaviours, and the internal regulatory variable responsible for activating the avoidance response and giving rise to our sense of disgust.

2. What to eat: expected value of consumption

As put forth by numerous theorists of disgust, including Darwin [20], Angyal [21], Tomkins [22] and Rozin et al. [4], food (and water) consumption is perhaps the most prominent behavioural domain in which disgust functions. In broad terms, the disgust response in the domain of consumption is an output of programmes that evolved to evaluate matter along a single, critical dimension: whether such matter, if consumed, would promote or subvert organismic order [7,23]. Only a relatively small subset of matter in the world contains the accessible high-energy chemical bonds and nutrients needed to help the organism resist the constant tug of entropy. The remaining much larger subset of matter would, if consumed, tend to foster disorder, damaging the organism and decreasing the chances of survival and reproduction.

In general, there are three categories of threat that could cause an increase in the entropy of living systems: mechanical, chemical and biological. Mechanical threats detected in the mouth—threats such as barbs, thorns, extreme temperatures and other forces capable of inflicting direct tissue damage—stimulate nociceptor cells of the peripheral nervous system and ultimately trigger the perception of pain [24]. Chemical threats, specifically plant-borne toxins and biological threats—that is, pathogenic microorganisms—are assessed via circuits in the central nervous system and, when present, contribute to our sense of disgust. These latter two categories are our focus here.

Not surprisingly, humans are not unique in the need to evaluate potential ‘food' items. All animals perform this function, and, arguably, so too do all life forms. For animals in general, we suggest that the information-processing architecture tasked with assessing matter for possible ingestion is engineered to detect foods high in order-promoting qualities—sugars, salts and proteins—as well as foods high in order-disrupting qualities—plant toxins and pathogenic microorganisms. Because each category of fitness-promoting matter and fitness-jeopardizing matter is evinced via different cues, there are probably specialized detection systems that estimate the probability that each type of substance is present. These distinct estimates are computed from multiple environmental cues, and are then combined with moderating contextual factors such as nutritional state, health, experience and life stage to generate an internal regulatory variable we term the expected value of consumption. Below we discuss several inputs to the system computing an expected value of consumption, including several moderating contextual factors.

(a). Cues to pathogen presence and plant toxins

Environmental information bearing on the probable presence of pathogens on food can derive from any of the five senses. Vision enables pathogen detection at a distance, and thus plays a key role in avoiding foods that pose a potential threat. Visual cues to pathogen exposure include the presence of certain animals indicating the advanced decomposition of organic matter—animals such as flies, maggots and worms [5]. Additional visual cues to pathogens include the presence of excreta such as faeces, vomit, blood, mucus and pus—all established sources of multiple infectious diseases [3]. Indeed, disgust theorists have long posited that the prospect of consuming the waste products of animal bodies—particularly faeces—is the core elicitor of disgust [21,25].

Gustatory cues to pathogen presence include the ammonia-rich and hydrogen-ion-rich by-products generated by bacteria, which are detected by specialized taste buds and give rise to our sense of sourness [26]. Bacteria and other microorganisms often also leave an olfactory signature: when bacteria decompose dead animal matter, they release a variety of volatile, foul-smelling compounds, including thiols, ammonium and hydrogen sulfide gas—compounds that humans can detect even in minute concentrations (e.g. see [27]). Other senses rely upon evidence that is perhaps less direct but that nonetheless correlates reliably enough with pathogen presence. The tactile correlates of pathogen presence include moisture and sliminess—indicators of the warm and wet micro-conditions in which pathogens thrive [28]. Indeed, chimpanzees will avoid consuming bananas when the banana slices are touching wet, mushy dough as opposed to a dry, solid rope [29]. Auditory cues have not generated much empirical attention, but the sound of retching readily elicits disgust and nausea, and some evidence suggests that there may be cross-cultural regularities in verbal expressions of disgust [30].

In contrast with pathogens, plant toxins provide relatively few reliable correlates discernible through vision, olfaction, hearing or touch. Many plant toxins, however, consist of alkaloids, chemical compounds that bind to specialized receptors in the mouth and produce the taste of bitterness. The general lack of direct cues to the toxicity of food (apart from bitter-tasting alkaloids) may promote human reliance upon socially transmitted information regarding which plant species may be safely consumed. Evidence that humans possess specialized systems for reasoning about food safety comes from infants. Liberman et al. [31] found that infants expect other individuals to reject a food item after a target showed disgust at consuming the item, regardless of social group membership, a dimension that often bounds the attribution of shared attitudes. Thus, learning systems appear sensitive to encoding information regarding palatability, a design feature that would have been especially helpful for discerning toxic from non-toxic plant matter.

(b). Contextual factors

The model of consumption we are proposing includes separate streams that estimate the probability of pathogen, toxin, sugar, salt and protein presence, all of which are taken as input by a compiler, one that estimates an expected value of consumption. But the assessment of pathogen presence should not necessarily lead to all-out consumption avoidance under all circumstances, just as the presence of sugar should not lead to uninhibited consumption under all circumstances. Our consumption psychology instead exhibits flexibility, but non-random flexibility that derives from the different factors the system was designed to take as input [8,32]. Prior research into the acquisition of taste aversions and preferences identifies a number of such contextual factors. Here, we highlight three: nutritional state, prior experience and developmental stage.

(i). Nutritional state

The nutritional state of the organism can shift the costs of error associated with food assessment. The costs of failing to detect pathogens when they are in fact present will generally outweigh the costs of inferring the presence of pathogens when they are in fact absent [5]. Under normal circumstances, forgoing a few calories matters little: hunger is a better option than death. But the calculus of food assessment changes dramatically when starvation or dehydration threatens: the costs of forgoing nutrition climb, while the relative costs of possible pathogen presence decline. As a result, internal representations of a person's nutritional state should feed into calculations of expected consumption value, increasing consumption value—and thus reducing disgust—even when cues to pathogens are present, or when cues to sugar, salt and protein are lacking. We note that this shifting of costs probably does not apply to plant toxins—the potency of plant-based poisons (e.g. strychnine and atropine) generally does not allow for potential error. This invariance could be engineered by ensuring that elevated probabilities of toxin presence are not muted by information regarding nutritional status, whereas elevated probabilities of pathogen presence are.

Research is consistent with the notion that nutritional state influences the disgust response. Thirsty men, for instance, found the odour of fermented fish less disgusting than did well-hydrated men [33]. Likewise, men deprived of food displayed less of a facial disgust reaction towards mouldy corn than did non-deprived men [34]. Note that when these men were in a depleted state, disgust decreased in response to microorganism presence. But disgust does not appear to decrease in response to toxin presence. For instance, Kauffman et al. [35] found that hungry participants were less likely to consume quinine-containing milkshakes, and Stevenson et al. [36] found that thirsty participants were less likely to drink quinine-containing liquids. When the body is in a depleted nutritional state, the consumption system appears to weight cues to toxin presence even more negatively, which is what bitter-tasting quinine is perceived to be. So whereas the consumption system relaxes its objection to items harbouring microorganisms, probably in a very controlled manner, it intensifies its objection to toxins. Better to risk illness from which one can recover, than death.

(ii). Prior experience

An individual's previous experience with candidate foods plays a key role in that individual's taste preferences, their food aversions and the emergence of a disgust response in a specific situation [4,37]. One key experience that shapes future estimates of consumption value is nausea. Indeed, classic research [38,39] suggests that the information-processing architecture underlying food preferences and aversions is predisposed to link nausea to foods consumed hours earlier, even when such foods are consciously known not to be the actual cause [40,41]. This predisposition is a causal bet and reflects the mind's assumption that nausea felt currently derives from food consumed earlier, and not from the weather, the sound of your neighbours arguing or the television programme you are watching. And given that the linkage of specific foods to nausea results in powerful aversions that endure long after the original experience, these predispositions probably reflect an evolved inferential structure targeted towards the threats posed by pathogens and toxins, where the costs of failed detection are high.

One way in which learning could be engineered is if there exists a prior value of consumption produced for each item ingested. When someone experiences nausea after consuming an item, this value is updated in a negative manner and influences future decisions accordingly. When one is met with no negative consequences after consuming an item, this variable is updated to be positive, promoting consumption in the future. Food neophobia and pickiness in eating probably reflect the operation of this system (e.g. see [42]).

(iii). Developmental stage

The risks and costs of exposure to pathogens and toxins vary as the organism develops and matures. As a result, different developmental stages should be associated with variation in the level and nature of defence against these threats. Specifically, we might expect to see increased disgust-mediated aversion to types of foods associated with specific toxin-based or pathogen-based threats during the life-history stages when those threats are greatest. One stage that has garnered substantial research attention is that of pregnancy, particularly the first trimester when the fetus is more vulnerable to the teratogenic effects of toxins and when the mother's suppressed immunocompetence increases her vulnerability to infection, especially from food-borne pathogens [43]. Consistent with this overarching hypothesis [44–46], women during the first trimester of pregnancy exhibit greater disgust sensitivity than do women in the second and third trimesters, particularly towards various food items [47].

With respect to engineering, sensitivities to toxins and pathogens can be regulated by changing how strongly various inputs affect the expected value of consumption. During pregnancy, for instance, estimated probabilities of toxin presence could have a much larger effect on consumption behaviour, raising the threshold for what makes a particular food item acceptable. And in children, after they begin weaning and sampling foods from the social environment, toxin sensitivities might remain elevated due to the continued threat of consuming plant-borne toxins [45]. Thus, adaptive flexibility can be achieved via the tinkering of how different inputs and contextual factors influence the computed expected value of consumption.

(c). Integration: the expected value of consumption

According to our model, the estimated probabilities of pathogen, toxin, sugar, salt and amino acid presence are combined and traded off against various contextual factors by a dedicated compiler that outputs an internal regulatory variable: expected value of consumption. The expected value of consumption of a given item can range from very low to very high. When high, the expected value of consumption will motivate approach, ingestion and perhaps a sense of deliciousness. Disgust, we argue, is merely the felt response when a low expected value of consumption is produced; indeed, the subjective experience of disgust, including strong motivation to avoid consumption of the substance, may be viewed simply as the felt magnitude of this variable (cf. [5]). The intensity of disgust will vary depending on all of the inputs and contextual factors discussed above. And hence meat with maggots on the fork and extremely bitter compounds on the tongue might trigger the most intense of disgusts, whereas piles of dung in the distance might trigger far less disgust. (Our model even predicts that disgust should be triggered in response to rocks or leaves offered as food. It should not (initially) be the intense disgust one feels towards maggots or vomit, but put some rocks and leaves up to your friend's mouth—put it to their lips and have them open their mouths for a taste—we suggest you'd be met with intense disgust).

Because low expected values of consumption can result from any combination of inputs or contextual factors (e.g. elevated probabilities of toxin presence or pathogen presence, large current pathogen load (i.e. cues of immune system activation are present), cues of pregnancy or current nutritional status), our model does not distinguish between ‘distaste' and ‘disgust' proper, in contrast with some prominent theories of disgust (e.g. [4]). Yet we do maintain that estimated probabilities of toxin presence and pathogen presence might have different effects on our system. While both will lead to low expected values of consumption, they might trigger separate response cascades, particularly physiological cascades. For instance, an elevated estimate of toxin presence might trigger sweating, vomiting and water consumption, whereas an elevated estimate of internal pathogen presence might activate components of the immune system.

Importantly, and germane to the other articles in this issue, many other animals probably possess a system that computes expected values of consumption to guide what to consume and what to avoid. Whether they feel disgust is another matter, but it is plausible that other animals possess circuitry that performs functions similar to those described above. While the issue is perhaps more difficult to explore in non-humans, evidence suggests that internal calculations about what is beneficial to consume and thus sought after change in systematic ways, reflecting a dynamically updatable system. For instance, animals that are ill (particularly the great apes) have been known to self-medicate and seek out particular bitter-tasting plants to consume [16,48–50]. To the extent that plant secondary compounds aid in the fight against bacterial or viral infections, consuming them might have conferred a substantial fitness advantage [51]. This is an explanation for why this behaviour evolved. One explanation for how this behaviour occurs in real time is that internal detection of infection alters the weighting of ‘toxin presence' when computing an expected value of consumption, specifically by weighting bitterness of plants more positively. In conjunction, an adaptive design would also include retrieval of stored information regarding the kinds of plant matter one witnessed others consuming (and not dying from). In this way, consumption psychology dynamically addresses the current need state in a culturally relevant and fitness-promoting manner.

3. What to touch: expected value of contact

The second domain in which disgust operates is contact. From a pathogen's point of view, one route of transmission from host to host occurs when hosts make physical contact. Furthermore, because pathogens tend to be well adapted to a particular host species [14], from the host's point of view, conspecifics tend to pose the greatest threat of horizontal transmission. As reviewed in Tybur et al. [5], the mouth, anus and genitals are three pathogen portals. Additionally, our skin serves as both a landing and launch pad for invasive microorganisms. We should thus expect to see heightened sensitivities to contacting others in these locations, particularly when there are signs of pathogen presence. Indeed, many animal species avoid contact with conspecifics who show signs of infection [16]. For instance, ants remove dead colony members from their nest [52]; the spiny lobster, despite being a highly social animal, will refuse to share dens with lobsters showing signs of disease [53]; killifish avoid other killifish showing signs of disease [54]; and mandrills will avoid grooming group members infected with parasites [55]. In general, there is a remarkable array of behaviours that have evolved to mitigate the costs associated with contacting contaminated individuals.

But beyond direct contact with conspecifics, pathogens can also be transmitted indirectly (e.g. via aerosol [15]). Remaining outside the sneeze or cough zone would have prevented contracting airborne pathogens. The environment also poses contact threats, perhaps best conceptualized as indirect effects of conspecific presence. It would have been beneficial, for instance, to avoid contact with faeces and resting or sleeping sites colonized by growing populations of pathogens such as mould, fungi, worms and bacteria. Various species provide evidence of these kinds of behaviours [56]. Baboons alternate their sleeping sites, presumably to let sites ‘air out' [57]; monkeys have dedicated defecation sites to avoid parasites [58]; sheep avoid grazing on grass contaminated by faeces [59]; rainbow trout swim away from parasitic worms that cause blindness [60]; and even the worm Caenorhabditis elegans avoids contact with bacteria placed in its Petri dish [61]. Across the animal kingdom, researchers have identified a variety of behaviours linked with avoiding contact with substances harbouring disease-causing agents.

(a). Cues to pathogen presence

In humans, for avoidance behaviours to occur, perceptual systems need first to detect cues associated with pathogen presence. Probably, many of the cues to pathogen presence for the purpose of avoiding contact are similar to those used for the purpose of avoiding consumption. Visual features associated with sores, lesions, bodily fluids and faeces would have cued surfaces and substances associated with an increased chance of disease transmission. Not surprisingly, humans and a number of other animals tend to avoid conspecifics with such features [62]. For instance, Curtis et al. [63] found that images depicting disease risks such as biological fluids—bloody, pus-like fluid—and open, wet skin lesions elicited far more disgust than non-biological fluids such as blue goo and lesions that had scabbed over. They also found that an image of a man with damp hair and reddish skin blotches was rated as more disgusting than a man depicted with dry hair and no skin blotches. Although it is tough to say whether subjects were responding to the wet versus dry hair and/or the presence of a skin infection, results are consistent with the notion that humans are sensitive to cues indicative of a body fighting an infection (e.g. fever and inflammation).

Other animals too serve as vectors for disease transmission [14–16]. Hence, decisions regarding which surfaces are safe to touch should be sensitive to cues reliably correlated with pathogen threat. For instance, the presence of ecto-parasites—assessed either from visual identification of insects and worms or from visible cues indicating their presence, such as rashes, bumps and lesions—should be taken as input to decisions on what counts as a pathogen threat and is thus safe to touch. Likewise, avoidance of animals such as rats, flies, cockroaches and other animals that dine on detritus or of animals showing cues to infection would have helped to avoid potential contamination [1]. As discussed above, many animal species are sensitive to other animals and conspecifics exhibiting signs of disease or serving as potential vectors of contamination (for a review, see [16]).

In addition to visual and olfactory cues to assess threats of contamination, humans and non-humans also use tactile cues to determine whether surfaces are safe for contact. Because pathogens tend to flourish in warm and wet environments and on animate beings, substances that share these properties should elicit greater avoidance. And they do. As mentioned above, Oum et al. [28] found that subjects rate dough as more disgusting than solid rope, a difference that becomes more pronounced when the dough is wet. In a similar fashion, Sarabian et al. [29] found that chimpanzees recoiled if their hand unexpectedly touched a piece of dough as they reached for a banana slice. Chimpanzees were also less likely to eat pieces of banana left atop mushy dough when compared with pieces of banana left atop a piece of rope. Generally speaking, humans and chimpanzees thus appear to show the same type of behaviour in response to a substance potentially harbouring pathogens.

In sum, we suspect that, in the human psychological architecture (and probably in the architecture of many other species), there exists a system that takes as input cues associated with the presence of microorganisms on conspecifics, other animals and surfaces in general, and then computes for a given surface the probability of pathogen presence. But should the detection of cues to pathogens lead to all-out aversion regardless of the context? Certainly not. Probably, there are particular contextual factors against which avoidance behaviours are traded off in favour of alternative fitness-promoting behaviours.

(b). Contextual factors influencing contact decisions

Much like how our food system weighs information regarding pathogen presence against other factors such as nutritional state in decisions regarding what to consume, our contact evaluation system also regulates touch by considering factors beyond cues to pathogen presence. We do not avoid contact with everybody all of the time. In fact, there are good reasons to expect to see trade-offs in decisions about whom to touch and when.

(i). Genetic relatedness

The first context in which we expect to see decisions to trade off risks of contamination in favour of contact is genetic relatedness. The benefits of assisting kin (e.g. one's offspring) often would have outweighed the costs associated with infection. As shown by Case et al. [64] and Stevenson & Repacholi [65], disgust differs when pathogen cues emanate from different groups of people. By and large, individuals identified as kin are less likely to evoke disgust when it comes to physical contact. Despite the very same physical properties, our brain appears to use information regarding relatedness to regulate our perceptions of the costs versus benefits of contact.

(ii). Relationships of mutual value

But because humans form cooperative relationships with non-kin (e.g. friends and mates), kinship could not be the only factor regulating contact. In non-human primates, friends groom one another, a behaviour that performs multiple functions but presumably evolved to rid individuals of external parasites [66]. Likewise, in humans, friends often engage in behaviours bringing them into close physical contact, suggesting that closeness, however this is represented in the brain, is another factor taken as input by systems regulating decisions regarding contact (for instance, see [67]).

(iii). Sexual arousal

The last factor we raise is sexual arousal. If other people represent planets of pathogens just waiting to leap from host to host, how in the world does sex occur? That is, if we have a pathogen-detection psychology that identifies other people's bodily fluids as potentially infectious and that motivates avoidance, how can two people overcome this intense contact aversion to engage in sex? What is needed is an override to the contact avoidance system in response to a (desirable) mating opportunity. Thus, another input to the system assessing the suitability of contact is what Tybur et al. [5] term expected sexual value, a variable that captures the expected fitness consequences of selecting a particular individual as a sexual partner. As we describe in the next section, this variable itself is a composite of a number of features that ancestrally were critical in evaluating a particular individual as a sexual partner. In general, however, when someone is perceived to be a very attractive potential sexual partner, that high sexual value should decrease the perceived costs associated with contact, facilitating sexual reproduction.

One prediction, then, is that sexual arousal should mitigate contact disgust [68]. There is good evidence that this is indeed the case. For instance, Stevenson et al. [69] found that men who viewed sexually arousing images exhibited reduced disgust towards sex-related stimuli (e.g. touching lubricated condoms). Similarly, Borg & De Jong [70] found that women who were sexually aroused reported less disgust associated with sexual tasks (e.g. adding lubricant to a vibrator).

The system works in reverse, too. Supporting the idea that disgust and sexual arousal exist along a continuum, not only does sexual arousal dampen disgust, but disgust also reduces sexual arousal. Fleischman et al. [71] found that women primed with disgust before being shown an erotic video reported less sexual arousal than did women in a control condition. The effect of disgust on sexual arousal could not be explained as being due to a general emotional arousal, as women who were primed with fear before watching an erotic video had the same levels of arousal as in the control condition.

In sum, the avoidance response in terms of whom to touch and when is expected to be flexible, taking into account dimensions that would have influenced survival and reproduction in ancestral environments. Three such dimensions include genetic relatedness, social closeness and mating context. However, for these dimensions to result in adaptive behaviour, they need to be integrated and traded off against each other in a manner that tends to have positive fitness outcomes.

(c). Integration: expected value of contact

Adaptively regulating contact behaviour—in humans and non-humans alike—requires a compiler that integrates the relevant information bearing on the problem of what and who to touch under a given set of circumstances. We posit a second internal regulatory variable, an expected value of contact, that performs this function, at least in humans and presumably also in other primates and animal species. When elevated (there are few cues to pathogen presence; the person is of high sexual value (low kinship) or a person is a close relative (low sexual value)), the expected value of contact motivates approach and contact; when low (e.g. pathogen cues are present), it motivates avoidance and our subjective sense of disgust.

Genetic relatedness, as mentioned above, is one factor that modulates the contact avoidance response. Given interactions involving a high degree of relatedness, cues to pathogen presence, such as a running nose, diarrhea and vomit, do not decrease the expected value of contact as they would in the absence of relatedness. Kinship suspends or at least dampens contact disgust—and only contact disgust.

The fact that kinship has a very different effect in the sexual domain, where it intensifies disgust, suggests there is a separate system governing sexual contact. That is, disgust, as a response designed to guard against the communication of pathogens via contact, does not automatically buy you sexual avoidance. It would seem that the same variable, an estimate of genetic relatedness, is taken as input by another system, one governing mate choice, the third system to which disgust is linked.

4. With whom to mate: expected sexual value

The last internal regulatory variable linked to disgust is expected sexual value, an index that guides decisions regarding whom to pursue, accept and avoid as a sexual partner for oneself [5]. As with expected values of consumption and contact, we suspect that systems that compute an expected sexual value evolved long before humans walked the Earth. Other animals too had to evaluate potential mates and faced similar adaptive problems when doing so. No doubt differences in neural circuitry exist between species, but we suggest that embedded in many species is a system that evaluates other individuals and motivates pursuit or avoidance in a manner that tended to enhance fitness over the course of that species' history. Here, we posit what this system might look like in humans and our nearest relatives.

(a). Cues to sexual value

There are two main components of sexual value, each with dedicated detection systems and compilers: systems for assessing mate value and systems for assessing relatedness.

(i). Mate value

The first component of sexual value includes the qualities that have typically been discussed in the evolutionary literature as constituting a person's mate value [72,73]. Reproductive potential and health are two main criteria that contribute to female mate value, whereas health and ability and willingness to obtain and share resources are main criteria of male mate value.

However, critically, assessments of mate value do not buy sexual attraction. After all, men and women are equally capable of assessing the mate value of both men and women [74,75], yet they are typically sexually attracted to individuals of only one sex. In addition, we are capable of assessing the physical attractiveness of our relatives, yet we typically do not experience intense sexual arousal towards kin. This is why we should expect to see an intermediate compiler that adaptively weights and trades off the many factors that contribute to assessments of the fitness outcomes of selecting an individual as a sexual partner.

(ii). Kinship

The second main component of sexual value, kinship, is a dimension assessed via the presence of cues that reliably correlate with an individual being a type of close genetic relative in ancestral environments. As reviewed elsewhere [7,76,77], there are a variety of parameters the mind takes as input when computing an estimate of relatedness for a given individual. Depending on the cues available, a kinship estimator generates an estimate of relatedness for each individual encountered, which then guides motivations to avoid an individual in sexual contexts [77,78].

Together, the mate value of a given individual and that individual's degree of relatedness are two critical inputs to a system that assesses sexual value. Mental systems that assess sexual value appear to take as input the mate value of both men and women. And the different weightings of a male's and a female's mate value in conjunction with the different factors influencing how mate value gets traded off against kinship lead to a range of sexual values and, hence, sexual preferences and behaviour. One way a sexual value system could be engineered is as follows. In the mind of a heterosexual male, all things equal, male mate value might be negatively weighted, whereas female mate value is weighted positively, varying according to assessments of her fertility and health. This can help explain why, for heterosexual males, reactions to females tend to be more on the lustful side of the continuum, whereas reactions to males tilt more towards the disgust side of the continuum—all else equal, females tend to post higher expected sexual values than do males. Likewise, in the mind of a heterosexual female, female mate value is probably weighted negatively, whereas male mate value will depend on the presence of cues associated with resource acquisition, status, intelligence, kindness and the other dimensions shown to influence assessments of male mate value.

For both men and women, however, an elevated estimate of kinship will override (under most circumstances) any effects of positive mate value on sexual value. That is, by virtue of having been categorized as kin, a male of high mate value will register as having low value as a sexual partner for a given female. Likewise, for a male, a female who is categorized as a sister might have a high mate value, but given a context in which other mates are available, the mind will generate a very negative sexual value causing intense sexual disgust when such behaviours are considered. Indeed, there is a large body of literature revealing that, in humans and non-humans alike, individuals avoid mating with individuals categorized as kin, especially when other mating opportunities are available [79–82].

(b). Contextual factors influencing mating decisions

As much as assessments of mate value and kinship can explain, however, they do not explain the wide variety of sexual preferences and aversions we observe across individuals and within an individual over time. Indeed, there are several contextual factors that would have influenced the success of mating decisions in ancestral environments. Here we briefly address three.

(i). Gender

Gender is one important factor that influences the calibration of systems guiding sexual partner choice. The organizational and activational effects of hormones in the developing fetus and during puberty lead to differences in the circuitries guiding mate choice in men and women. The major selection pressure that led to these differences is the different level of minimum parental investment required on the part of men and women to rear young [83]. Males, generally speaking, can increase their reproductive success by taking advantage of sexual opportunities with successive fertile women. But not so for females, who are limited by their obligatory investment in gestation and lactation. This difference is reflected in the tuning of psychological programmes that assign sexual value. For instance, the Coolidge Effect and studies by Clark & Hatfield [84] suggest that males more readily assign high sexual values to new potential sexual partners than do females. Given their greater costs of reproduction, it probably takes females longer to apply a given positive magnitude of sexual value to a potential mate, when compared with males. Overall, one's own gender and the manner in which one's psychology was calibrated during development and puberty play a critical role in how sexual value is assigned to others in the environment.

(ii). One's own mate value

Recent studies on mate preferences lend support to the idea that one's own mate value also factors into sexual motivations. For instance, Morgan & Kisley [85] manipulated males' sense of their own attractiveness via feedback regarding whether others believed them to be of high or low market value and then presented them with attractive and unattractive faces of women. Examination of attention as measured via event-related brain potentials revealed that all males attended to attractive faces, regardless of whether they were given feedback that they were of high or low market value. Attention differed, however, for unattractive faces: males told that they were of low market value spent significantly more time attending to unattractive faces when compared with males told they were of high market value. Similarly, Lee et al. [86] found in a study examining meet-up requests and attractiveness ratings on an online dating website that less attractive individuals were less selective in whom they agreed to go on dates with, whereas more attractive individuals were more selective. Last, Yong & Li [87] had both men and women hold a small sum (approx. $84) or large sum (approx. $2100) of money and found that the men—but not the women—who held the large sum of money increased their minimum requirements in the attractiveness of a potential date. But women are not immune—they too show sex-specific effects of self-assessed mate value. Buss & Shackelford [88] found that more physically attractive women held higher standards for an acceptable mate in terms of his masculinity, income, desire for a home and children, and devotion as a partner. Together, these results suggest that an internal variable that captures a sense of one's own mate value modulates estimates of sexual value and motivations to pursue another as a sexual partner.

(iii). Availability of potential mates

A last contextual factor predicted to influence sexual value estimates is mate availability. One way to engineer flexibility into the mate choice system is to enable the recalibration of others' sexual attractiveness based on the available pools of mates. Put another way, design features that caused an individual to modulate their assessed sexual value of another based on the potential to acquire partners of even higher sexual value would have tended to leave more offspring in ancestral environments when compared with design features that set, for example, an elevated minimum threshold and left it there throughout adulthood.

A general prediction that flows from this view is that the more restricted the perceived mating pool, the more relaxed one's preferences will become. And, depending on how limited one might be, one might even begin to find kin, same-sex friends or even other animals as potential sexual outlets. To our evolved mind, some sex might be perceived as better than no sex. For men who are incarcerated, the limited pool of potential mates turns other men into sexual outlets. Somehow, a variable assessing the range of opportunities available modulates the magnitude of expected sexual value, perhaps by changing how mate values are weighted in the calculation of sexual value. However, because the costs of reproduction are lower for men, we should expect to see greater risk-taking sexual proclivities in men when compared with women. That is, women are likely to remain resistant to engaging in sexual behaviours that potentially jeopardize the health and viability of offspring far longer than men.

Evidence regarding the effect of a restricted mating pool comes from data on Taiwanese minor marriages, a form of marriage common in the sixteenth century in which a young infant girl was adopted into her future husband's family and reared alongside him throughout childhood [89]. Though typically used as an example of the Westermarck effect (because childhood association led to reduced fertility and greater rates of divorce and extramarital affairs), Taiwanese minor marriages illustrate that when one's mating pool is severely restricted, one might even marry, have sex with and raise children with individuals categorized as kin. Further evidence that mate availability influences sexual partner choice comes from the area of familial sexual abuse. As discussed by Finkelhor [90], secluded living arrangements (e.g. in rural communities) contribute to the risk of sexual abuse within the family.

(c). Integration: expected sexual value

Another person's mate value, their degree of relatedness, one's own gender and mate value, and availability of potential mates combine to generate a third internal regulatory variable, an expected sexual value for a given individual. When elevated, expected sexual values generate a sense of sexual attractiveness and motivate sexual pursuit. When low, expected sexual values cause avoidance behaviours and give rise to our sense of disgust. We postulate that expected sexual values, much like the expected values of consumption and contact, exist along a continuum. For decisions regarding sexual partner choice, the absence of an elevated expected sexual value does not automatically translate into disgust. That is, just because you aren't sexually attracted to someone, it does not mean you are intensely sexually disgusted by them. You might be lukewarm, suggesting that the expected sexual value is low, but perhaps not low enough to trigger disgust. One prediction is that, again, much like decisions regarding consumption and contact, in the sexual domain, one cannot feel both disgusted and attracted at the same time towards the same object.

5. The three regulatory variables causing disgust: where's morality?

Together, expected values of consumption, expected values of contact and expected sexual values govern decisions regarding what to eat, what/who to touch and who to select as a sexual partner (table 1). All three domains of behaviours are intimately linked to pathogens, either because they include responses to prevent horizontal pathogen transmission between hosts or vertical transmission of pathogens down the generations. The same—or at least similar—sense of disgust governs decisions when these regulatory variables are of very low magnitudes. Refinements over evolutionary time might have led to some differences in response to decisions regarding consumption, contact and copulation, but as studies have shown, there is overlap in motivational responses and even neurological hardware (e.g. [91]).

Table 1.

Summary of inputs and contextual factors that contribute to the production of three internal regulatory variables that, when low, generate our sense of disgust. We do not consider the list of contextual factors (or inputs, for that matter) to be exhaustive; to explain the terrific flexibility of our consumptive, contact and sexual behaviours, probably there are other variables that influence the strength of our motivations.

	inputs	contextual factors	internal regulatory variable (IRV)	IRV output range
1. consumption	cues indicative of: sugars, salts, amino acids (order-promoting) microorganisms, plant toxins (order-jeopardizing)	nutritional status health status life stage prior experience and information	expected value of consumption	extremely low (disgust; avoid) to extremely high (delicious; devour)
2. contact	cues indicative of microorganisms	kinship other source of welfare valuation sexual value prior experience and information	expected value of contact	extremely low (disgust; avoid) to extremely high (desire; longing to be near)
3. copulation	cues indicative of mate value: health (microorganisms) fertility resource status	kinship own mate value mate availability prior experience and information	expected sexual value	extremely low (disgust; avoid) to extremely high (lust; longing to be near)

We suspect that many non-human animals share the systems we have outlined above, of course with species-specific refinements. We do not feel that disgust is unique to humans, at least not in terms of the information-processing systems that govern our felt sense of disgust. In this way, the model we outline above departs from previous models that suggest humans are unique in their disgust response [4,92,93]. However, much like Paul Rozin, Jonathan Haidt and colleagues [4,25], as well as Darwin [20] and Angyal [21], we suggest that an important domain of disgust concerns oral incorporation. Our model illustrates how plant toxins and pathogenic microorganisms are detected and evaluated to generate low expected values of consumption. We have concepts of bitter and sour by virtue of the detection systems for these compounds. Both, we argue, can generate disgust. Thus, there is no meaningful difference between distaste and disgust in our model.

Lastly, while there is substantial overlap, our model also departs from prior evolutionary treatments of disgust. In alignment with our model, other theorists argue that disgust is characterized by a primary function: to defend against pathogens, parasites and infectious disease [1–3,5,6,63,94–96]. We differ from a strict pathogen avoidance theory (e.g. [1]) in that we posit that inputs—other than estimated probabilities of pathogen presence—that generate low expected values of consumption can also elicit disgust. Though pathogens will be a main factor causing disgust in the domain of consumption, it will not be the only factor. We also differ, more or less, from previous evolutionary-inspired models in positing the existence of a separate sexual domain of disgust or in excluding morality as a proper domain of disgust (e.g. see [97,98]). Others, too, have called into question the existence of a separate moral disgust, citing evidence that the ‘disgust' reported in response to socio-moral violations may be largely metaphorical in nature [99–102] without necessarily generating the qualia and physiological components that characterize pathogen-related disgust.

As detailed elsewhere [7], we suggest that when expected values of consumption, contact and sex are low, that is, when disgust is triggered, these variables can be used by ‘moral' systems (i) to identify interpersonal harms (e.g. individuals who engage in behaviours that jeopardize reproduction or transmit pathogens to you, your kin or associates—for instance they sneeze on you; serve you food that's been dropped on the ground; they sexually molest you); and (ii) to identify potential behaviours to condemn as a means of catalysing a coalitional action (e.g. felt disgust at homosexuality is used as an input to decisions of whether to target, condemn and exploit homosexuals). However, one need not feel disgust to use the language of disgust. To the extent that disgust guides the formation of coalitions for the purpose of exploiting groups of lower formidability, another role disgust plays is to (iii) signal allegiance to a (typically leverage-positive) group. Thus, the trappings of disgust—the language and facial expressions—can be used to signal one's allegiance, either during the initiation or formation of norms (e.g. the banning of foie gras) or, once norms have taken full effect in the larger population (e.g. codified in law), to protect against group condemnation and exploitation (see [103–107]).

6. Conclusion

Disgust is an emotion that evolved to oversee protection against the selection pressures posed by pathogens. Considering the manner in which pathogens exploit their hosts, we have identified three routes of transmission—consumption, contact and copulation. Whereas our consumption and contact psychologies help to avoid the horizontal transmission of pathogens, our sexual psychology helps to avoid the vertical transmission of pathogens and the effects they have on offspring survival and future reproduction. Our model identifies three internal regulatory variables that oversee behaviour in each domain. Clarification of how disgust performs its various functions in the domains of consumption, contact and sex also allows for the generation of hypotheses regarding how disgust is used in the social domain to generate and enforce norms with the goal of either protecting against harm or identifying groups to exploit.

No doubt, this is not the last model of disgust. However, we hope future models will take seriously the task of specifying, in information-processing terms, the kinds of systems that would have evolved to solve the myriad problems posed by pathogens.

Data accessibility

This article has no additional data.

Competing interests

We declare we have no competing interests.

Funding

We received no funding for this study.