A valence-free definition of sociality as any violation of inter-individual independence

Abstract

Well over 90% of studies in biomedical sciences are performed on single animals. While knowledge of the genetics, development and physiology of single individuals accrues, an understanding of the biological mechanisms by which individuals interact has barely budged. Yet many of society's greatest problems derive from an inability of humans to get along with each other. Studies in social neuroscience are primarily observational and rarely employ subjects who physically interact. Thus, social interaction represents a largely unexplored frontier of biology. The neuroscience that underlies social behaviour and interactions can and should be studied using the scientific method. However, a workable and objective definitional framework of sociality is needed for scientific progress in this field. Here we propose a definition that uses a test of independence from the presence of others. The null hypothesis is that a behaviour is independent from the influence of others. Rejection of this null hypothesis means that the actions of an individual depend on the actions of one or more other individuals. This definition has the advantages of not being contaminated by moral judgements or biases in favour of pro-social behaviour, and of being applicable to a wide range of physiological processes. The definition of a social behaviour proposed here says nothing regarding the valence of the behaviour with respect to others. Thus, a behaviour that is influenced by the presence of others may benefit, harm, or have no effect on others. It is hoped that this definitional framework for sociality will facilitate our understanding of the origins and mechanisms of social behaviour among animals including humans as well as offer efficacious approaches to social disorders such as autism.

1. Introduction

Humans generally consider themselves social. We believe ourselves to be so attuned to sociality, such natural experts, that we are seduced into accepting the accuracy of our own assessments of behaviours, situations, and species as social or non-social. This seduction is dangerous and allows common assumptions and beliefs to serve in place of dispassionate definitions. Here, we propose a novel way to frame and define sociality. We aim to provide an objective method to answer questions such as: what makes a species or a behaviour social?; how is a social activity different from a non-social activity?; or can one act socially when alone or non-socially when in a crowd?

2. The human as an animal

An inherent danger when using an everyday, non-technical term in science is that the flexible, blurrily demarcated everyday meaning clouds the precise definition that scientific endeavours need. ‘Social’ is one such word. Its commonplace usage in our life (‘social network’, ‘socializing’, ‘social person’ and so on) gives it a meaning that is as salient to human language users as it is ill-defined. As problematic as the vague definition of social is when applied to human-related activities, it is just as problematic as we move farther and farther away from the twig upon which Homo sapiens sits within the tree of evolution. What would constitute a social behaviour in a sponge, for instance? Note that whether sponges actually have social behaviours is irrelevant here; the point is that before judging whether they are social or not, one must define the word. Plainly, sponges cannot communicate with gestures or express jealousy, typical ‘social’ behaviours that we see in select primates. We therefore need to develop a definition for ‘social behaviour’ that satisfies two criteria. First, it must be applicable across the phylogenetic tree; second, it must allow for tests, conceptual or experimental, that can be used to determine whether a certain behaviour or process is social.

A definition of sociality that can be applied across the animal kingdom allows the lessons learned from social interactions of non-human animals to inform the human condition, to tell us truths that are not obvious from or may even be obscured by human reports, a sort of inverse anthropomorphism. It is likely that there are lessons to be learned. Microbes influence the mating behaviour, and thus the evolutionary course, of host animals [1]. Mammalian behaviour can also be driven by the microbes within as exemplified by aggressive dogs infected with rabies and toxoplasmosis-infected rats who approach, rather than run from, cats [2–3]. While our understanding of microbial influences on bacteria and invertebrate animal behaviour is progressing, the influence of the microbiome on human behaviour is just beginning to be explored.

We may be in a position to learn a great deal from non-human animals. It is sobering to think that just as syphilis was considered a mental illness in the nineteenth century, mood disorders that we now consider as neuropsychiatric may in fact be eventually revealed as the logical outcome of a particular microbe within the affected individual's microbiome. If we are to learn from sea stars, mosquitoes, wasps and other non-human animals, a definition of sociality should not require language-based conceptual cognition. A language-independent definition would also allow evaluation of the sociality of behaviours that have no human correlate—think jam-avoiding response in electric fish—or physiological processes for which there is no conscious awareness—think gastrointestinal peristalsis or cardiac output.

Enormous benefits accrue from studying the human as an animal. For nearly a century, the modern field of psychology has been hammering on the robust finding that human report is inaccurate [4]. People are not trying to deceive. Rather, humans are not privy to the inner workings of the brain through introspection. In a nice example of this, Brian Knutson and colleagues showed that the activity in the nucleus accumbens of 30 subjects was a better predictor of public crowdfunding choices than was the collective behaviour of the subjects [5]. In other words, the subjects made errors in deciphering the noisy signal from their own brain; their brains were truer to their intentions than was their behaviour. Of course, we know that non-human animals also make errors, for example, as they perform categorization near a category boundary. However, the Knutson example tells us something about our own ability to know what we like, our ability to know our own mind in an ecological setting. Furthermore, it suggests that decision-making in non-human animals—should a rat eat a morsel or hoard it—also results from noisy brain signals. By studying humans using the same methods as one would use on rats or monkeys, one can find truths that are not camouflaged by the ‘just-so’ stories that humans create. Moreover, the utility of human language opens up a large range of processes and questions that can only be addressed when methodology from studies of non-human animals is applied to humans.

3. Should rats or humans serve as the standard?

Empathy is an instructive case study for the bidirectional synergy between human and non-human studies. The existence of empathy among humans is commonly inferred from actions such as verbal report, consolation behaviour, donating money and helping another in distress. Rats perform similar actions including grooming another who has been shocked, interpreted as consolation, and releasing a trapped rat which the senior author and others have argued constitutes a rodent version of human empathy [6–7]. However, it is plausible that humans express a human version of the rat motivation to help another in distress. In other words, our construction of empathy may be a verbal smokescreen for actions that are actually based on the same brain activity that occurs in a rat to produce similar-appearing actions.

The reification of empathy depends on human self-report and self-reflection. Human empathy is a feeling that is no more visible from a starting point of human behaviour than it is from watching rats. The key difference is our language. A two by two table of the presence of empathy in rat (or any other non-human animal) and human yields four possibilities, only two of which are worthy of consideration. One popular notion holds that humans experience empathy and rats do not. The competing idea is the congruent condition in which rats and humans both experience empathy. This is not to say that rat and human experiences are the same any more than it is to say that a person's experiences stay the same through time or that different people experience the world the same by virtue of their membership in the Homo sapiens species.

4. Empathy is a morally neutral term

Empathy research is laden with unjustified assignments of positive valence. Certainly, empathy can motivate helping behaviours. However, empathy is ultimately a neutral term as it refers to the communication of affect and experience between two individuals. Understanding another's distress can certainly lead to efforts to reduce that distress but can also motivate efforts to exacerbate and maximize another's distress. For example, knowing another's state of mind can be used to act in a way that will be particularly psychologically injurious to a specific person, ‘personalized torture’ if you will.

Empathy is widely lauded as desirable by politicians, activists, celebrities and the general public. In nominating Sonya Sotomayor for the Supreme Court of the United States, President Barack Obama explained that he saw the ‘quality of empathy… as an essential ingredient for arriving at just decisions and outcomes.’ Websites with names such as cultureofempathy.com, rootsofempathy.org and compassionsociety.org promote empathy as a modern snake oil to combat a variety of societal ills and break down distrust between people. Putting aside the hyperbole that empathy can cure the social ailments of modern society, Paul Bloom has argued, in Against Empathy, that empathy produces more moral harm than good [8].

The trolley car paradigm illustrates the idea that empathy is not necessarily a moral or pro-social motivator. Consider that a railway car is heading straight for five people tied to the railroad track. You are standing next to a switch that will allow you to divert the train over to a track upon which only one person is tied. This dilemma asks whether it is better to act so that one person instead of five will be killed. Many people believe that reducing the death toll from five to one is the moral action. Now consider that the railway car is barrelling down a track toward a distressed stranger yelling for help whereas five unconscious strangers are tied down on the second track. In this scenario, emotional empathy for the single emoting person would favour pulling the switch so that five strangers die and one emotive person lives. Thus, empathy can drive acts that run counter to moral endpoints. As Bloom [8], writes, ‘empathy is a spotlight focusing on certain people, [leaving us] blind … to the suffering of those we do not or cannot empathize with…. It is innumerate, favoring the one over the many.’ Here, we argue that an inordinate focus on valence has retarded scientific progress in social neuroscience as it has in empathy research.

5. The term social is contaminated by valence

With empathy research in mind, it is instructive to consider social neuroscience from a sociocultural perspective. The evaluation of sociality has frequently been contaminated by valence, particularly by an unjustified focus on pro-sociality. We put forth a neutral definition of sociality that is less susceptible to judgemental bias because it evaluates a quality that is orthogonal to the effect of the behaviour on others. In other words, sociality is neutral with respect to whether the action in question leads to another's benefit in either the short or long term.

Typical topics of social neuroscience include empathy, cooperation, pair-bonding and the like. These topics are nearly universally viewed as desirable. Matusall and colleagues [9] write that, ‘today's characterization … of the term ‘social’ often sets priorities on ‘positive’ issues like cooperation, empathy, care,’ and so on, leading to a ‘contamination’ of biology with societal ideals.

As an example of contamination between the common usages of ‘social’ and ‘pro-social’, consider the definitions of social in the Merriam-Webster Dictionary: (i) ‘tending to form cooperative and interdependent relationships with others’; (ii) ‘living and breeding in more or less organized communities’. The latter applies particularly to insects and the former would appear to be widely applicable to the animal kingdom. These definitions belie the common bias that sociality means pro-sociality. However, territoriality, aggression and bullying are all social in that they are actions that require at least two actors. They cannot occur in worlds of single individuals. There is no justification for carving out the positive end of the interactive spectrum as social and excluding the more neutral and frankly antagonistic range of interactive behaviours.

6. A good definition of sociality is one that can be tested

Considerations and discussions of definitions often appear belaboured and more semantic than revelatory. Yet, definitions have enormous power over our assumptions and therefore the framing of our questions. The difference between sensation and perception, widely appreciated by those in social science disciplines and less so by biologists, is an example of a powerfully influential rubric. It is only with the understanding that perception is not a feature defined by an external stimulus that we can then ask, for example, how language influences perceived experience and thought [10–11]. Likewise, only with an accurate delineation of sociality can we start investigating, with any degree of rigour, the neural substrates of social behaviours.

7. Valence is not an innate property of behaviours

Risking a full digression into the nature of moralities, we find it necessary to discuss how valence is assigned to social behaviours. We often develop intuitive mappings between social behaviours and valence—exacting pain on others is bad while treating the pain of others is good. Yet valence is not an innate property of behaviours, but one relative to the context—the intention, the intended target, the unintended consequences, the environment, as well as the judger's perspective.

The baby that receives a vaccination does not experience the needle's assault as positive but the parents, and eventually the baby when she grows into a healthy adult, will be glad to have escaped the scourge of disease. We contextualize our own actions, distinguishing between assault and benevolent medicine. Helping a terminally ill spouse die is viewed as a murder by some and as a kind act by others. Euthanasia for a beloved pet with a terminal condition is nearly always viewed positively. Thus, valence is not easily assigned to human actions. This difficulty is only amplified when we consider the ant or shark or skunk, cases where we are unlikely to understand the full consequences of the actions performed by these animals.

Even pro-social actions may not serve uniformly pro-social ends. Consider an emergency room physician who treats a young man with a badly cut arm. This act would be considered pro-social. Now consider that that young man injured his arm when throwing a punch at his wife who ducked just in time to send his arm through a window. Now the ultimate effect of treating the young man takes on a different hue, increasing the likelihood that he will be able to try again and ultimately succeed in injuring, possibly even killing, his wife. If the young man does indeed go on to kill his wife, does this outcome diminish the positive nature of the physician's initial action of treating the arm injury? Or think of one person giving another a gift. What could be clearer than this being a pro-social act? Yet, sometimes receivers do not want the gifts of givers, a feeling that is captured by the Japanese phrase arigata-meiwaku, meaning the feeling ‘when someone has done you a favour that you didn't want from them, and which may have [even] caused you difficulty, but you're required to feel grateful anyway’ [12].

Scenarios where the actual effect does not match the intended effect highlight a major problem with classifying the valence of social actions. Additionally, the focus on positive or negative valence when applied to social behaviour is laden with cultural implications and moral judgement, which hinders rather than helps objective scientific progress.

8. A neutral approach to sociality

We define sociality as any significant deviation from behaviour exhibited by an individual when that individual is within a group of two or more. By defining social behaviours as those that lack independence from the influence of others, we can construct quantitative rubrics by which to categorize behaviours or processes as non-social (independent), under-social (less likely to happen than would be expected by independence), or above-social (more likely to happen than would be expected by independence). The under- and above-social terms simply refer to whether the probability of an action occurring is less or more than would be expected if individuals within a group acted independently (figure 1).

Figure 1.

The black lines represent the predicted probability of an action occurring (y-axis) in a group of independently acting individuals. Both curves assume that all individuals perform the target action with the same frequency when alone but can be modified to accommodate individual variation. Observed probabilities are under-social (red areas) if the probability is less than would be predicted by a group of independently acting individuals. In contrast, actions that occur more frequently than would be predicted by a group of independently acting individuals are termed above-social (blue areas). (a) The null prediction is given for a group of three as the likelihood that one individual acts is varied (x-axis). (b) The null probability of an action with an individual frequency of 0.10 increases as group size (x-axis) increases.

For example, consider that there is a 10% chance that any one individual will wave to a passing sailboat. The probability increases to 34% for four independently acting individuals and becomes highly likely (81%) in a lakeshore crowd of 16 independent actors. The null hypothesis is that groups of people are independent actors. Rejection of the null hypothesis due to a significant deviation from independence is evidence for sociality. In this scenario, the chance that not one of 64 people will wave at a sailboat is roughly one in a thousand. About as unlikely is that three independent individuals, each with a 10% probability of waving, will all wave at a sailboat. Thus, both coordinated waving and a consistent failure to wave can qualify as social within appropriately sized groups. Coordinated waving is not more social than the uniform failure to wave. Both involve individuals acting differently when in a group than when alone.

Several advantages accrue from this definition. It is objective and the idea is extraordinarily simple. The question posed asks whether individuals within a group are acting independently or not. Acting independently is the null hypothesis. If the data do not allow for rejection of the null hypothesis, then there is no evidence that individuals are not acting independently, and thus not socially. If on the other hand, the null hypothesis is rejected, then the actions of an individual are not independent of the actions of one or more other individuals. They are acting socially in one way or another.

The simple analysis presented here yields a categorical answer that a given process or behaviour is either social or not and if it is social, the direction (under or above) is revealed.

9. The range of potential processes that can be evaluated for sociality

The test for sociality above concerns an action, which can be treated as a Boolean variable: did an action occur or did it not occur? Yet not all behaviours or physiological processes, any one of which should be testable using our definition of social, can be evaluated using this particular approach. For example, a different analytical tactic is needed to determine if the timing of eating bouts is influenced by the presence of others. Still different methods are needed to evaluate whether an autonomic process such as the rate of intestinal peristalsis or pupil size is socially influenced. Still different approaches are needed to analyse the independence of location (figure 2). Regardless of the process of interest, the proposed definition of sociality boils down to comparing the variable in question in individuals that are alone to those same variables when the individuals are part of a group. It is worth noting that such an approach avoids the primarily observational methods used in social neuroscience studies to date, employing representations of others while a metric such as self-report or fMRI image is collected [13]. Few studies use subjects who physically interact. Thus, these studies ignore the engagement of one individual with another and the ensuing two-way information processing.

Figure 2.

Cumulative position histograms of 32 rats in a 50 × 50 cm arena. All rats were tracked for the first 10 min of the session. (a) Rats placed alone in the arena spent most of their time in or near the corners. (b) Rats placed in the arena with a rat trapped within a Plexiglas restrainer (white dotted line) spent most of their time around the restrainer, particularly near the restrainer door (left side of restrainer).

A recent example of sociality comes from David Eilam's laboratory [14]. A large arena of more than 30 m² contained 16 pieces of food laid out in a central grid pattern. Rats learned to collect all food pieces when placed solo in the arena. Rats were then tested in pairs or in groups of three. An obvious possibility would be that the individual rats within a group would divide up, each collecting food independently with the result that the total time needed to collect all the baits would be less than the time required by a solo rat. Thus, the null hypothesis here is that the total time needed by multiple rats is approximately the average total time needed by a single rat divided by the number of rats. However, this result did not occur. Instead, the time needed was greatly increased over the predicted value. Rats in the group condition stayed together, travelling to each food location in tandem, even though this decreased the amount of food acquired by some rats. As is evident in the available video and as described by the authors, ‘rats are observed to be more preoccupied in socializing than in collecting the baits.’ Because of the inefficiency of travelling en masse to each of the food baits, there was no difference in the total time needed to collect all the food pieces between rats tested alone or in groups. This is a powerful example of a social influence on behaviour. Note that in this particular example, our definition of sociality escapes the conundrum of having to decide whether rats are ‘cooperating’ or ‘competing’, or in fact giving it any valence at all.

The number of animals may not exert a linear effect on behaviour. In particular, there may be a discontinuity between an individual and a group of two or more. Two, three, or 10 extra animals may provide no more ‘buffering’ from stress than does one additional animal [15]. Both step and graded effects are of interest and both represent social influences, albeit of different varieties.

Another nonlinear influence on behaviour may occur when the peak effect occurs with a moderately sized group. Groups that are either larger or smaller may exert less of an effect on behaviour, producing a U-shaped (or inverted U-shaped) curve. To return to the example of stress, it is reasonable to conjecture that over-crowding produces stress much as isolation does so that animals exhibit minimal stress markers with a moderately sized group.

In sum, to clearly describe a social influence, a ‘social dose–response curve’ with a range of group sizes is needed. If that curve deviates in any fashion from the expected curve of independence, then the process is socially influenced.

10. Do non-social animals or activities exist?

If the influence of others on an individual's actions is the mark of a social behaviour, then what would constitute a non-social action? One candidate for a non-social action involves the great philosopher Immanuel Kant, who led a highly structured and rigid life. It is said that precisely at 16.30 every day, Kant walked up and down his street exactly eight times [16]. If this account is accurate, then Kant walked regardless of who was or was not present at his home and also without regard to the people who may or may not have been on the street. Such an action would represent an act that is independent of others and therefore quintessentially a non-social act. Kant's rigid, non-social afternoon walk sounds contrived and unbelievable, a reaction that reflects the likely apocryphal nature of the story.

Regardless of whether Kant participated in a non-social daily walk, it is likely that non-social actions are infrequent. Yet non-social processes certainly occur. For example, nightly release of growth hormone is an excellent candidate for a possible non-social process.

11. The range of potentially social behaviours

A walk through a public park reveals a wide panoply of behaviours and highlights the advantages of a clear and objective definition of sociality. One popular way to define sociality is by location. Groups of animals that herd, flock, or school together are considered social whereas animals that live and hunt solo for long stretches of time are considered non-social. Now consider a public park on a beautiful sunny weekend day. From a distance, hundreds of people may be present in 10–15 degrees of the visual field. They appear as a crowd, a herd. However, upon closer inspection, one may see a person sitting alone, drawing a tree; couples sitting near to each other with each of them reading silently; groups chatting as they pass a bottle of wine around to share with each other; a juggler practising his trade.

By any definition of the word, the group that is actively talking and sharing wine is engaging in a social activity. How about the couple quietly reading in close proximity? Would each member of the couple be at the park if she or he were without the other? How about the lone person who is sitting and drawing or juggling? Why are these solo individuals drawing or juggling at the park? Why not in the privacy of their home? Could it be that they have come to the park to see and hear others or to be seen and heard by others? And what of the individual that sits alone at home on that same sunny weekend day? She or he may be basing her decision to not go to the park on the presence of other people rather than on a fear of skin cancer or the difficulty in reaching the park. In other words, it is possible that a desire not to socially interact is the main motivating factor for a solo individual to stay home from the park.

Staying home in order to avoid others is fully as social as is joining a group at the park. The hermit's likelihood of going to the park is inversely related to the number of people at the park. The partygoer is acting in a way that seeks out others and the hermit in a way that avoids others, both motivated by a consideration of others.

12. The bystander effect

The bystander effect is an interesting example of a social effect. The term refers to the paradoxical finding that the likelihood of a group of people helping is less than the likelihood of one individual helping and far below the likelihood that a group of independently acting individuals will help [17]. Typically, the actions of subjects tested alone are compared to those of subjects tested with confederates, research staff who were instructed not to perform the action in question. Since the confederates do not help, the question becomes whether subjects studied in groups are more or less likely to help than subjects studied alone. Consistently, subjects are far more likely to help when alone than when in the presence of confederates. Thus, the presence of confederates diminishes rates of helping, an example of an under-social response.

The under-social effect of bystanders is mediated by an effect on the perceptions of the subjects [17]. While the method of communication remains unknown, confederates communicated their perception sufficiently well that confederates' perceptions trumped the subjects' own perception. The same effect has been observed in rodents. Rats who have an aversion to a food will eat that food if exposed to other rats who are eating it [18]. While the ultimate effect of the bystander effect is group conformity, the process appears to represent an integration of the perceptions of others with those of the self. In the end, the perceptions of the others can override the perceptions of the self.

13. What advantages accrue from defining sociality?

The benefits of clearly defining sociality will be of particular use when it comes to understanding conditions such as autism spectrum disorder (ASD). Distinguishing an individual who avoids social contact from one who is indifferent to social contact is of critical importance. The former recognizes others, which in turn motivates avoidance behaviour. In this case, therapeutic initiatives then need to focus on why cues that typically elicit affiliative behaviour instead trigger avoidance in affected individuals. In contrast, the non-social individual fails either to recognize others or to use the presence of others as motivation for action. Distinguishing between these two possibilities then becomes a necessary first step to an understanding of pathophysiology that can lead to the development of treatment strategies. It is expected that there are forms of ASD that fit into each of these categories.

Another realm in which a clear definition of sociality is an advantage is in the understanding of the evolutionary origins of empathy. The commonly held dominant idea is that empathy and empathy-related social behaviour evolved from parental care for offspring. Yet this idea has not been rigorously tested. An alternative idea is that social behaviour evolved to solve ecological problems faced by particular species within their own environmental niche. Using a quantitative definition of social behaviour, these competing ideas are testable by examining closely related species that occupy different niches and use different reproductive strategies.

Ethics

All procedures conformed to the highest ethical standards and were reviewed and approved by the University of Chicago Institutional Animal Care and Use Committee.

Data accessibility

The data used in figure 2 are available upon request from the senior author (pmason@uchicago.edu).

Authors' contributions

P.M. drafted the paper and produced figure 1. H.S. produced figure 2. H.S. and P.M. revised the paper together extensively.

Competing interests

We declare we have no competing interests.

Funding

We received no funding for this study.