Encyclopedia of Evolutionary Psychological Science

Living Edition
| Editors: Todd K. Shackelford, Viviana A. Weekes-Shackelford

High-Cost Altruistic Helping

  • Hannes RuschEmail author
Living reference work entry
DOI: https://doi.org/10.1007/978-3-319-16999-6_1556-1


Prosocial Behavior Inclusive Fitness Dictator Game Reciprocal Altruism Direct Fitness 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.



In acts of high-cost altruistic helping, substantial risks to the own well-being or considerable losses of social and/or economic capital are voluntarily incurred in order to benefit others.


Thomas H. McCann was one of the first recipients of the Carnegie Medal for civil heroism. On June 29, 1904, McCann jumped from a bridge 25 ft high into Portland Harbor and dragged Alfonso Sekosky, an 8-year-old boy, to safety on a boat, but was too fatigued to get into it himself and was drowned. McCann’s case is quite representative: since its inception in 1904, the Carnegie Medal has been awarded to almost 10,000 “individuals in the US and Canada who risked their lives to an extraordinary degree saving or attempting to save the lives of others” (carnegiehero.org, September 2016). About 20% of the recipients of the Carnegie Medal did not survive their attempts to help others (Becker and Eagly 2004).

Why is it that ordinary people like McCann, who was only 32 years old at the time of his death, voluntarily take such tremendous risks to their own well-being in order to help others? Or, asking from the biological perspective, how can such prosocial behavioral inclinations be conserved or even be promoted by evolutionary forces, if they substantially increase the chance that their bearers eliminate themselves from the human lineage?

This entry reviews potential answers to these questions in light of the available empirical evidence. After a brief clarification of terms, the first main section focuses on theories that can explain the preservation and propagation of prosocial behavioral traits that cause high costs to their bearers. The second section discusses the problems that research on high-cost helping faces, before the third and fourth sections review the evidence available from controlled experimental studies and the analysis of documented historical cases of high-stakes prosocial behavior in the domains of civil life and warfare. The concluding section points out important open questions and suggests directions for future research.


Dying in the course of his heroic action, Thomas H. McCann paid a high price, his own life, for the protection of a boy who, presumably, was a total stranger to him. But what if he had died rescuing one of his sons instead? Would we think of his action as equally honorable? If we take the eligibility criteria of the Carnegie Medal as a proxy for prevalent societal norms about which kinds of actions are considered heroic, then the answer is “yes”: “Persons not eligible for awards are: […] members of the immediate family, except in cases of outstanding heroism where the rescuer loses his or her life or is severely injured” (carnegiehero.org). Still, callous Darwinian evolutionary logic states that the price paid by McCann is higher, ceteris paribus, than the one a father pays when dying for the life of his offspring. Obviously, thus, the Darwinian costs of high-stakes prosocial behavior can differ. Therefore, before we review potential explanations for how such behaviors might evolve, let us briefly recapitulate the underlying evolutionary calculus.

The currency of natural selection is fitness, which can be approximated as the expected number of copies of a given gene that an individual is able to successfully pass to the next generation. Therefore, when analyzing the prospects of evolutionary success of a certain behavioral trait coded for by one or multiple genes, we need to take into account that trait’s direct fitness consequences on its bearer, as well as its indirect fitness consequences on individuals bearing the same gene or genes. Furthermore, it is important to note that the fitness balance, F, of a certain trait can only be evaluated after aggregation over the entire lifetimes of all individuals affected by the behavior, because both direct and indirect fitness consequences, D and I, respectively, can each have an immediate and a delayed component. For short, let us define the fitness balance of a trait as: F = D now  + I now  + D later  + I later . As we will see, all of these components can play crucial roles in explanations of the evolution of high-cost prosocial behaviors and altruism, which, in terms of direct and indirect fitness consequences, can now be defined as follows.

According to West et al., four types of behavioral traits need to be carefully distinguished, based on their direct fitness consequences (i.e., D now  + D later ) for the acting individual: mutualistic “+/+” behaviors benefit all affected parties, spiteful “–/–” behaviors cause costs for all parties. selfish “+/–” behaviors benefit the acting individual at a cost to those affected, and altruistic “–/+” behaviors benefit those affected at a cost to the actor (West et al. 2007). While natural selection can potentially favor mutualistic and selfish behaviors straightforwardly, because of their positive effects on direct fitness, spiteful and altruistic behaviors can evolve only if their negative effects on direct fitness are offset by sufficiently high indirect fitness benefits, i.e., if their inclusive fitness balance is positive.

One important point to note about this definition of altruism is that not only the immediate direct fitness consequences of a behavior need to represent a cost to the acting individual but that their overall direct fitness effects need to be negative, i.e., D now  + D later  < 0. Thus, if the immediate direct costs of behaving to the benefit of others are offset by direct benefits accrued later, such behavior does not fall under the category of altruism sensu West et al. but represents mutualistic behavior instead. To avoid terminological confusion, thus, we will refer to any kind of action to the benefit of others as “helping” and only speak of altruism if D now  + D later  < 0 actually holds for a specific action.

A second important point to note is that the effective costs of a certain type of prosocial behavior can differ between individuals. Obviously, the risk of drowning while swimming to the rescue of a person in distress is much lower for an experienced and athletic swimmer than for someone who swims only infrequently. That means: for the latter D now  < 0 likely holds, while the experienced swimmer is more likely to face a D now  ≈ 0 type of risk. The classification of an action as “costly,” thus, requires a thorough prior evaluation of the fitness effects an action potentially has on the respective acting individual. As will be discussed later, this second point mainly is a problem for empirical studies on helping behavior. The theories reviewed in the following section, however, can all explain how helping might evolve also when it is effectively costly, i.e., when D now  < 0.

Theory: How Can Helping Evolve, Particularly if It Is Extremely Costly?

Numerous theories exist that can explain how behavioral traits causing immediate direct fitness losses to their bearers might eventually be preserved and even favored by natural selection (see, e.g., Kurzban et al. 2015; Nowak 2012, for overviews). Using the terminology just introduced, their common rationale can be summarized as follows. In order for such a trait not to go extinct in the long run, its inclusive fitness balance, F, must be zero or positive. If one or more of the terms of this sum are negative, thus, at least one of the other terms must be positive and large enough to compensate for the negative terms.

All theories on the evolution of costly helping behavior start from the assumption that D now  < 0. They differ in their assumptions about which other components of the inclusive fitness balance are eventually compensating for these costs (see Table 1). The first set of theories – costly signaling, direct reciprocity, and indirect reciprocity – all identify delayed direct fitness benefits to the helper, D later , as the compensating factor. Costly helping of this kind, thus, does not represent altruism sensu West et al. (2007). Let us briefly review these theories in turn.
Table 1

Overview of potential explanations for the evolution of helping behavior classified by their immediate direct, D now ; delayed direct, D later ; immediate indirect, I now ; and delayed indirect, I later , fitness consequences


D now

D later

I now

I later


Costs compensated by

Costly signaling





D later + I later

Direct reciprocity





D later

Indirect reciprocity





D later

Inclusive reciprocity




I later

Kin selection




I now + I later




Costly signaling. The main idea behind the theory of costly signaling is that individuals differ in characteristics that determine their attractiveness to others as social and/or sexual partners but that are not directly or reliably observable (Spence 1973; Zahavi 1975). By engaging in costly helping, those individuals who can afford the respective costs, e.g., because they are physically fit or wealthy enough to do so, can reliably signal two characteristics that increase their attractiveness: their physical or economic capabilities and their prosociality. Such unforgeable public displays, in turn, increase the chances for the signalers to benefit from social interaction with choosey high-quality partners later on, by whom they might otherwise have been ignored (for further discussion, see, e.g., Barclay 2011).

Direct and indirect reciprocity. While the recipients of help eventually play no role other than causing a cost to the signaler in costly signaling theory, they are more important for the theories of direct and indirect reciprocity (Nowak and Sigmund 2005; Trivers 1971). Direct reciprocity theory assumes that the initial cost of helping is eventually overcompensated by the returns from a longer-term cooperative partnership between the helper and the recipients of help that results from this initial act of benevolence. Indirect reciprocity theory, on the other hand, assumes that helper and recipient might never meet again. Instead, the cost of helping here is compensated indirectly through help that the helper herself receives from third parties later on who base their decisions to help on reputation. In order to maintain a good reputation, and thus to receive help if needed, one must help others if they are in need. Hence, under regimes of indirect reciprocity, costly helping buys helpers access to a community of cooperative social partners and eventually makes them better off than living in a community of non-helpers (for further discussion, see, e.g., Roberts 2008).

As we have just seen, the three previous theories assume delayed, but positive, direct benefits to the helper, i.e., D later  > 0. One might be tempted to interpret this assumption as meaning that these theories only work in case helpers always survive, because otherwise they cannot benefit from these delayed direct fitness benefits. This is not the case, though. All that D later  > 0 requires is that there is a positive chance of surviving for the helper, because what counts from the evolutionary perspective is not the fate of any single individual, but rather the “average fate” of all bearers of a specific trait over a longer period of time, of course. Still, there are at least two theories that can potentially also explain the evolution of traits that lead to their bearers’ death with certainty, i.e., theories that work even if D later  < 0.

Inclusive reciprocity. The first of these theories, inclusive reciprocity, is an extension of the theory of indirect reciprocity (Albert and Rusch 2013). Its central idea is that reputation is not necessarily limited to individuals, but can span entire groups or families of individuals. If parental reputations are inherited by the offspring, e.g., altruistic behavioral traits can evolve that cause a parent’s death, if they lead to a sufficiently high increase of the offspring’s reputation, i.e., if they make I later sufficiently large. Note that under inclusive reciprocity, altruistic traits can evolve even if they do not benefit the offspring directly in any way and even if they cause immediate costs to them, because inclusive reciprocity works through delayed indirect benefits.

Kin selection. The other theory that is potentially able to explain the evolution of “lethal helping” is kin selection, i.e., inclusive fitness theory. Just like inclusive reciprocity, kin selection theory assumes that direct fitness costs to the helper are eventually compensated by indirect benefits, i.e., benefits to the helper’s kin. The main difference between the two theories is that inclusive fitness theory assumes that the act of helping itself causes these indirect benefits, while inclusive reciprocity assumes that they are mediated by the reputational gains that the act of helping causes for the offspring of the helper, i.e., by “family reputation.” Inclusive reciprocity, thus, requires that individuals are able to track reputations and to condition their behavior on them, while inclusive fitness theory is less demanding. Kin selection only requires that altruistic acts have positive indirect effects on helpers’ kin in the long run. Note that this does not even require that individuals are able to discriminate between kin and non-kin. All that is needed is a clustering of kin groups in a given population such that the average expected fitness consequences of helping are higher for relatives of the helper than for non-kin (Kümmerli et al. 2009).

Mismatch. Before we move on to discuss how the explanatory value of the theories just outlined could be tested on the background of the available evidence on acts of high-cost helping behavior, one important possibility must not go unnoticed. As is known, natural selection usually works slowly. Given that, from an evolutionary perspective, the time during which humans lived in small bands of quite closely related mobile hunter-gatherers is “just” over, contemporary high-cost altruistic helping, i.e., helping with D now  + D later  < 0, may have become dysfunctional through the rapid changes in human social organization over the last 15k years. Therefore, while such prosocial behavior may have had significant positive effects on kin in ancestral times, the “mismatch hypothesis” states that these effects are no longer present today and that evolution might simply not have caught up with this fact.

The Intricacies of Studying High-Cost Helping

It is not hard to see that testing the aforementioned theories against each other is not a straightforward task. While the mismatch hypothesis makes for a suitable null hypothesis, and may thus receive some support in case all alternative theories should fail, researchers trying to find supportive evidence for these alternative theories face numerous problems.

One intricacy on the theoretical side is, for example, that many of the suggested explanations for the evolution of high-cost helping are not mutually exclusive. As can be seen from Table 1, evidence of delayed direct benefits, D later  > 0, for instance, can yield support for costly signaling as well as for direct and indirect reciprocity. Disentangling further which, if any, of these three theories receives more support if evidence for D later > 0 is actually found, thus, requires refined research designs and/or more fine-grained data gathering efforts.

On the empirical side, on the other hand, researchers of high-cost helping behavior face even more taxing challenges. For obvious reasons they cannot put experimental subjects into life-threatening situations or confront them with decision situations involving other types of high personal costs. To work around this problem, at least three approaches to approximating such situations under controlled conditions have been used in the literature. Naturally, they all have their respective up- and downsides.

The first commonly used approach is to focus on the question of how – real or fictitious – persons who engaged in high-cost helping are perceived by third parties. Using this approach, researchers can experimentally manipulate certain characteristics of the helper, the recipients of help, the situational context, etc. The second approach is to ask subjects how they would behave in hypothetical scenarios requiring high-cost helping – the characteristics of which can again be experimentally manipulated, including the priming of subjects with mind-sets of interest, like mating motives or high group coherence. Of course, both these approaches face the common problem that the data they yield stem from self-reports and are not readily generalizable to the real-world analogs of the persons and situations studied in the laboratory. They are, however, able to establish causal relations, which makes them an indispensable tool in the study of high-cost helping.

Apart from using self-reports, researchers have also used economic decision-making experiments to study high-cost helping. In these, subjects are endowed with money and asked to decide if they would like to use their endowment to help one or multiple persons in need. In addition to being able to reveal causal relations as well, the second benefit of studies of this kind is that subjects make decisions that have real monetary consequences for themselves and other persons. Through increasing the size of the endowment, these studies can then approximate situations of helping at high personal costs. Apart from the practical challenge that such studies can become arbitrarily expensive, they face at least three additional problems. The first is that endowments usually are “windfall” money to the subjects, i.e., free money that they might spend more laxly than money they earned outside the laboratory. The second problem is that the same amount of money may be of quite different value for different subjects. Comparisons of economic decisions between individuals, or even countries and cultures, therefore, should ideally simultaneously control for individual- and group-level differences in wealth and in the endowments’ purchasing power. The third problem, finally, is fittingly captured in the commonplace that “money can’t buy everything,” meaning that even decisions involving very high monetary stakes may psychologically not have the same gravity as fitness relevant decision situations involving immediate threats to one’s survival and/or reproductive prospects.

Complementary to studying high-cost helping experimentally under controlled conditions, a second main branch of the literature has developed that analyzes data on documented historical cases of such behavior (also see Rusch 2016). Just like all other studies that are based on historical data, this branch of the literature faces all the problems caused by the impossibility to control the conditions from which these data originate, of course. These problems include, but are not limited to, sampling bias, undetected confounding factors, data availability, and measurement problems.

Sampling bias, for instance, represents a potential obstacle for researchers who gather case data from sources like the Carnegie Hero Fund. Naturally, those cases of civil heroism available from this source were selected for being publicly honored by the fund’s selection committee. The underlying process of nomination, screening, and eventual selection or rejection by this committee might, of course, be influenced by shifts in societal norms, changes in the composition of the committee, adjustments of the eligibility criteria or new views on how they are to be interpreted, etc.

Apart from potentially biasing selection committees’ work, societal norms can also represent confounding factors that may affect research using historical data, of course. The fact, e.g., that – to date – there is only one woman among the roughly 3,500 recipients of the highest US military award, the Medal of Honor, must not be taken to indicate that women are very unlikely to behave bravely in combat. Obviously, instead, women simply did not have the opportunity to display such bravery until the fourth quarter of the twentieth century during which they were gradually admitted to all branches of the US military. Other important potential confounders include contraceptives, which interfere with the study of reproductive success after the 1950s, or socioeconomic status and urbanization which may reduce the exposure of individuals to certain types of life-threatening situations.

In addition to being subject to these aforementioned complications, finally, research on historical cases of high-cost helping virtually always suffers from limited data availability. Trying to track the reproductive success of specific individuals, e.g., is a time-consuming endeavor and usually results in incomplete data. A standard problem in this process is the identification of extramarital offspring, which is usually impossible. Furthermore, the available information on specific individuals, even if accessible, often is limited, thus forcing researchers to either base their analyses on suboptimal proxy variables or even to leave interesting questions unanswered.

In summary, research on high-cost helping faces both theoretical and, maybe more importantly, numerous practical problems. Still, as the following sections will show, a number of instructive empirical studies on this type of behavior exist. They are listed in Table 2. Although each of these studies faces specific limitations, some of which were just outlined, a number of more general insights already become apparent from the overall picture drawn by the available evidence. Therefore, instead of describing each study individually, the following two sections try to summarize these more general insights.
Table 2

Synopsis of the existing literature on high-cost helping



Data source/s

Main findings (paraphrased)

Economic studies on voluntary helping

Engel (2011)

Meta-study, econ

Dictator game studies

Average share given: ≈28% (N = 616); Small negative effect of endowment size (N = 158)

Smeets et al. (2015)


Millionaires playing dictator games

Millionaires are much more generous in the dictator game than other samples

Karagözoğlu and Urhan (2016)

Meta-study, econ

Multiple game types

Negative effects of endowment size in multiple dictator game studies

Controlled laboratory studies on high-cost helping

Kelly and Dunbar (2001)

Rating exp.

Mixed sample (N = 120)

Females prefer risk-prone brave males to risk-averse non-brave males, and men are aware of this preference

Farthing (2005)

Rating exp.

Students, two exp. (N’s = 100//117)

Females and males preferred heroic physical risk takers as mates, preference stronger in females

Griskevicius et al. (2007)


Students, four experiments (N’s ≈ 200)

Experimentally induced romantic motives produce strategic and sex-specific displays of helping and consumption

Barclay (2010)

Rating exp.

Students (N = 175)

Altruists more desirable for long-term relationships than neutral individuals

Moore et al. (2013)

Rating exp.

Mixed sample (N = 77)

Reports of helping behavior increased the attractiveness of both men and women as potential long-term sexual partners

Rusch et al. (2015)

Rating exp.

Students, two experiments (N’s = 92//340)

Females regard men more sexually attractive if they are war heroes; no effect for male participants judging female war heroes

Studies analyzing historical case data on high-cost helping

Blake (1973)

Hist. data

Medal of Honor (N = 325)

Potential biases in selection process of Medal of Honor recipients are discussed

Blake and Butler (1976)

Hist. data

Medal of Honor (N = 207)

Two types of distinguished actions are identified: “war winning” and “soldier saving”; the former is most clearly associated with officers and the latter with low-ranking enlisted men

Blake (1978)

Hist. data

Medal of Honor (N = 207)

“Altruistic suicide” is found to be higher in more cohesive than in less cohesive groups and more likely among enlisted men than among officers

Riemer (1998)

Hist. data

Medal of Honor (N = 125)

Partial replication of Blake (1978)

Rusch (2013)

Hist. data

Medal of Honor (N = 966)

Altruistic behavior toward comrades seems to occur more frequently when soldiers are on the defensive

Rusch and Störmer (2015)

Hist. data

Medal of Honor (N = 988)

Descriptive information on 988 Medal of Honor recipients of WWI, WWII, Korea, and Vietnam

Rusch et al. (2015)

Hist. data

Medal of Honor (N = 123)

Proxies for reproductive success are compared between 449 regular veterans and 123 surviving Medal of Honor recipients of WWII; results suggest that the heroes sired more offspring than the regular veterans

Lay et al. (1974)

Hist. data

Carnegie Medal (N = 101)

Men directly intervene in emergencies more often than females; responsive bystanders tend to act alone; rural people are more heroic than urban people

Toronto Metropol. Police Civilian Citations (N = 147)

Johnson (1996)

Hist. data

Carnegie Medal (N = 676)

Detailed analysis of actions of Carnegie Medal recipients focusing on the relationship of rescuers with the recipients of help

Becker and Eagly (2004)

Hist. data

Carnegie Medal (N = 8,706)

Carnegie medalists are disproportionately men, but other actions yield representations of women that were at least equal to and in most cases higher than those of men

Righteous Among the Nations (N = 10,343)

Kidney donors (N > 60 k)

Peace Corps volunteers (N = 9,622)

Doctors of the World applicants (N = 76)

Lyons (2005)


UK Newpapers (N = 286)

Males were highly more likely to rescue than females, the typical rescuer was a low status male rescuing another male

Rand and Epstein (2014)


Carnegie Medal (N = 51)

Self-reports by Carnegie Medal recipients about their decision to act were judged to be overwhelmingly dominated by intuition

McNamee and Wesolik (2014)


Carnegie Medal (N = 30)

Carnegie Medal recipients’ parents’ expectation that they help others was the only parenting factor that differentiated the Carnegie Heroes from a random sample

Evidence from Controlled Experimental Studies on Helping Behavior

The standard paradigm for studying costly helping behavior in economic experiments is the dictator game. In this game, subjects are randomly matched into pairs. One subject is then assigned the role of the dictator and the other the role of the recipient. The dictator receives an endowment from the experimenter and can decide how much of it she would like to share with the recipient. The recipient, on the other hand, does not make any decisions. Experimental designs usually ensure that dictators can make their decision in private and that dictator and recipient do not know each other, cannot communicate, and are very unlikely to meet after the experiment. Generally, thus, the aim of dictator game designs is to make sure that dictators do not have to fear any negative consequences, but also cannot expect any positive ones, irrespective of what they decide. Still, across studies (Engel 2011) and across cultures (Henrich et al. 2001), remarkable proportions of dictators decide to share a part of their endowments with the recipient, 28% on average (Engel 2011). As discussed above, however, these endowments are usually comparably small.

Stakes effects in dictator games. What happens, now, when endowments are increased? Two recent meta-studies tried to tackle this question (Engel 2011; Karagözoğlu and Urhan 2016) and arrived at mixed conclusions. Analyzing data on average giving rates in 158 experimental conditions from studies in which endowment sizes were systematically varied, Engel (2011) finds that increasing endowment size seems to have a small negative effect on dictators’ willingness to share. However, for his larger sample of 440 conditions from dictator games for which dollar equivalents could be computed, Engel finds no such effect, with stakes ranging from $0 to $130. Karagözoğlu and Urhan (2016) arrive at similarly mixed conclusions. Intriguingly, a recent study on the giving behavior of dollar millionaires, furthermore, found millionaires playing with an endowment of €100 to make the most generous dictator game decisions observed in the literature so far (Smeets et al. 2015).

With respect to high-cost helping, thus, the existing literature on dictator games suggests two things. For one, the readiness to share windfall money with others may be relatively insensitive to stake size. From a methodological perspective, this could be regarded as good news, as it yields some support for the assumption that costly helping behavior can be studied in the lab using comparably small stakes without losing too much generalizability. Importantly, however, the range of stakes used in the existing dictator game literature is still limited to three-digit dollar amounts. We currently do not know what happens if stakes are increased beyond this (for further discussion, see Karagözoğlu and Urhan 2016). For the other, as the millionaire study by Smeets et al. (2015) indicates, it is important to keep in mind that the same monetary stakes may be of very different value for different subjects. Revealingly, when asked by Smeets et al. to explain their decisions, many of the millionaires gave answers like “With my money, 100 euro is easy to give” (see the supplements to Smeets et al. 2015).

Rating studies. A somewhat clearer picture emerges with respect to the results of studies that asked participants to rate fictitious persons who engaged in high-cost helping. Focusing predominantly on high-cost helping as an honest signal of desirable qualities as a mate, the existing studies consistently find positive effects of helping on ratings of attractiveness. In addition, Griskevicius et al. (2007) report that subjects who were experimentally manipulated into a mating mind-set self-reported a higher readiness to engage in certain types of costly helping behavior toward third parties. Thus, convergent evidence exists in the literature that costly helping does increase perceived attractiveness and that it may also be used, intentionally or unintentionally, as a means to display personal qualities on the mating market. The literature is less univocal, however, with respect to the question if using costly helping as an honest signal is a sex-specific phenomenon or not. While Griskevicius et al. (2007) report that their experimental manipulations yield stronger effects in men, many other studies find that both men and women positively react to displays of costly helping by rating helpers as more attractive than both non-helpers and neutral persons with otherwise identical characteristics. Furthermore, the literature also is inconclusive with respect to whether helpers are preferred as long-term or short-term partners or both. Unfortunately, because of the focus of most studies on costly helping as an honest signal in the context of mating, the literature also does not offer a decisive answer with respect to the question of whether there are additional benefits to costly helping through indirect reciprocity. At least, Kelly and Dunbar (2001) report that helpers are also rated as more preferable social partners, i.e., friends, by women.

In summary, the literature on costly helping as a “courtship display” (Barclay 2010) provides quite consistent evidence that various types of costly helping behavior actually seem to increase perceived mate value. However, the questions of whether this effect is sex-specific and whether costly helping is also an effective signal in other contexts than mating are not answered conclusively by the existing studies.

Evidence from Documented Historical Cases of High-Cost Helping

Given these results from controlled laboratory studies, the logical follow-up question of course becomes, if they can help us understand the occurrence of real-world high-cost helping behavior better. To date, however, only comparably few studies have tried to shed light on this question, each of them focusing on rather distinct aspects of this kind of behavior. Their data, furthermore, mainly stem from two sources only: the Carnegie Hero Fund and the official Medal of Honor citations (also see Rusch 2016).

Civil heroism. Recipients of the Carnegie Medal, “civil heroes” for short, have been analyzed in five studies so far (see Table 2). Their main common observation is that the Carnegie Medal is predominantly awarded to men (roughly 90% of all recipients, see Becker and Eagly 2004). Notably, the only study using an alternative data source, UK newspaper articles on cases of civil heroism (Lyons 2005), confirmed independently that civil heroism is more likely to be displayed by men. Two other observations were also made repeatedly. First, civil heroes likely have an increased probability of belonging to the working classes (Johnson 1996; Lyons 2005). Second, civil heroism seems to occur more often in rural regions as compared to urban ones (Johnson 1996; Lay et al. 1974).

It is tempting to take these three repeatedly made observations as evidence supporting two of the theories outlined above. The overrepresentation of men among civil heroes suggests that this type of high-cost helping might be a sex-specific costly signal. The tendency of civil heroes to have low socioeconomic status and the more frequent occurrence of civil heroism in rural areas might indicate that indirect reciprocity is playing an important role, as anonymity is likely to be lower and a good standing in one’s community likely to be more important in rural areas and in social strata in which mutual favors cannot be substituted through buying the same services.

However, with respect to sex bias, Becker and Eagly (2004) argue that the overrepresentation of men among civil heroes might be due to the physically challenging nature of those specific types of acts of helping conventionally subsumed in the category of “civil heroism.” In fact, Becker and Eagly are able to show that for other types of brave and potentially very costly actions, such as hiding Jews from the Nazis or donating a kidney, among others, no such bias exists – and even that women are overrepresented among some of these other groups of helpers. For the findings on social status and region, on the other hand, Lay et al. (1974) and Johnson (1996) point out that similar caution needs to be applied. It could be that those types of situations allowing for actions of the types most frequently honored as civil heroism are just more likely to occur in rural regions and in contexts that members of the lower social strata are more frequently exposed to.

War heroism. The actions of the recipients of the Medal of Honor, “war heroes” for short, represent the second source of historical case data on high-cost helping behavior frequently used in the literature. Seven studies exist that have studied selected characteristics of these men and the contexts of their heroic actions. However, the only finding that has been repeatedly reported across studies concerns cases of “heroic suicide,” i.e., the deliberate sacrifice of the own life by covering an explosive device in order to save the lives of fellow soldiers in close proximity. Analyzing partially overlapping data extracted from the official Medal of Honor citations from different wars, Blake (1978), Riemer (1998), and Rusch and Störmer (2015) all report that this kind of high-cost helping behavior is observed more frequently in lower-ranking soldiers and in members of the Marine Corps. Other noteworthy, but singular, findings reported in this literature include that surviving Medal of Honor recipients of WWII seem to have had increased reproductive success compared to other veterans of this war (Rusch et al. 2015) and that acts involving direct help to comrades are more frequently observed during defenses (Rusch 2013).

The finding of increased reproductive success of WWII Medal of Honor recipients apparently yields direct support to the idea that war heroism might be a costly signal favored by sexual selection. However, in addition to pending replication using data from other times, regions, and sources, numerous confounding factors could not be ruled out by Rusch et al. (2015), and their finding therefore needs to be interpreted with due caution. To name just one limitation, they could not rule out, for instance, that the material benefits associated with the Medal of Honor might have allowed its recipients to support a bigger family as compared to regular veterans.

With respect to the observation that enlisted men and members of elite troops seem more ready to “make the ultimate sacrifice,” the same caveat applies as with apparent status and provenance effects in civil heroes. It could simply be that enlisted men and elite troops are more frequently exposed to situations in which such a sacrifice can be made. However, data that allow controlling for this potential confounding factor currently are unavailable and, unfortunately, are likely to remain so as gathering them seems hardly feasible.


This entry has reviewed selected evolutionary theories that can potentially explain how behavioral traits causing their bearers to lend help to others can be sustained and also promoted by natural selection even when the acts of helping cause high costs for the helpers. Subsequently, the intricacies of testing these theories against the results of laboratory and archival data studies were discussed. Finally, the existing empirical literature on high-cost helping was summarized briefly.

Contrasting the broad spectrum of suggested theories to the sparseness of available empirical work on high-cost helping, it becomes immediately apparent that intensified research is needed. The only theoretical suggestion, for which more than a modicum of empirical support has been found so far, is costly signaling. Costly helping does indeed seem to increase the perceived attractiveness of helpers as mates and may even have resulted in increased reproductive success in some historical cases. This should not be taken to mean, though, that the other theoretical suggestions – direct, indirect, and inclusive reciprocity and kin selection – have received no support so far. They simply have not been tested in the study of high-cost helping, yet.

Given the continuously increasing availability of data sources on historical cases of high-cost helping from online databases and the steadily improving accessibility of biographical information on historical persons, significant progress in this area of research can be expected in the nearer future, though. Particularly, since instructive recent studies have already pioneered ways of accessing and analyzing such sources (Lyons 2005; Rusch et al. 2015).

Apart from the imperative need to replicate earlier findings using alternative data sources and better controls to rule out potentially confounding factors, important research questions to be tackled include the following: To which extent is costly signaling in the form of helping a sex-specific phenomenon? Apart from increased mate value, which direct and indirect benefits, if any, do high-cost helpers regularly obtain, and what are their effects on helpers’ fitness balances? And finally, do the available theoretical suggestions suffice to explain the occurrence of the dark flip side of “lethal helping” in the form of suicide attacks (Atran 2003)?



  1. Albert, M., & Rusch, H. (2013). Indirect reciprocity, golden opportunities for defection, and inclusive reputation. MAGKS Joint Discussion Paper Series in Economics, 2013-19. Google Scholar
  2. Atran, S. (2003). Genesis of suicide terrorism. Science (New York, N.Y.), 299(5612), 1534–1539. doi:10.1126/science.1078854.CrossRefGoogle Scholar
  3. Barclay, P. (2010). Altruism as a courtship display: Some effects of third-party generosity on audience perceptions. British Journal of Psychology, 101(Pt 1), 123–135. doi:10.1348/000712609X435733.CrossRefPubMedGoogle Scholar
  4. Barclay, P. (2011). Competitive helping increases with the size of biological markets and invades defection. Journal of Theoretical Biology, 281(1), 47–55. doi:10.1016/j.jtbi.2011.04.023.CrossRefPubMedGoogle Scholar
  5. Becker, S. W., & Eagly, A. H. (2004). The heroism of women and men. American Psychologist, 59(3), 163–178. doi:10.1037/0003-066X.59.3.163.CrossRefPubMedGoogle Scholar
  6. Blake, J. A. (1973). The congressional medal of honor in three wars. The Pacific Sociological Review, 16(2), 166–176. doi:10.2307/1388359.CrossRefGoogle Scholar
  7. Blake, J. A. (1978). Death by hand grenade: Altruistic suicide in combat. Suicide and Life-Threatening Behavior, 8(1), 46–59. doi:10.1111/j.1943-278X.1978.tb01084.x.PubMedGoogle Scholar
  8. Blake, J. A., & Butler, S. (1976). The medal of honor, combat orientations and latent role structure in the United States military. The Sociological Quarterly, 17(4), 561–567. doi:10.1111/j.1533-8525.1976.tb01723.x.CrossRefGoogle Scholar
  9. Engel, C. (2011). Dictator games: A meta study. Experimental Economics, 14(4), 583–610. doi:10.1007/s10683-011-9283-7.CrossRefGoogle Scholar
  10. Farthing, G. W. (2005). Attitudes toward heroic and nonheroic physical risk takers as mates and as friends. Evolution and Human Behavior, 26(2), 171–185. doi:10.1016/j.evolhumbehav.2004.08.004.CrossRefGoogle Scholar
  11. Griskevicius, V., Tybur, J. M., Sundie, J. M., Cialdini, R. B., Miller, G. F., & Kenrick, D. T. (2007). Blatant benevolence and conspicuous consumption: When romantic motives elicit strategic costly signals. Journal of Personality and Social Psychology, 93(1), 85–102. doi:10.1037/0022-3514.93.1.85.CrossRefPubMedGoogle Scholar
  12. Henrich, J., Boyd, R., Bowles, S., Camerer, C., Fehr, E., Gintis, H., & McElreath, R. (2001). In search of homo economicus: Behavioral experiments in 15 small-scale societies. American Economic Review, 91(2), 73–78.CrossRefGoogle Scholar
  13. Johnson, R. C. (1996). Attributes of Carnegie medalists performing acts of heroism and of the recipients of these acts. Ethology and Sociobiology, 17(5), 355–362. doi:10.1016/S0162-3095(96)00059-3.CrossRefGoogle Scholar
  14. Karagözoğlu, E., & Urhan, Ü. B. (2016). The effect of stake size in experimental bargaining and distribution games: A survey. Group Decision and Negotiation. doi:10.1007/s10726-016-9490-x.Google Scholar
  15. Kelly, S., & Dunbar, R. I. M. (2001). Who dares wins: Heroism versus altruism in women’s mate choice. Human Nature, 12(2), 89–105. doi:10.1007/s12110-001-1018-6.CrossRefPubMedGoogle Scholar
  16. Kümmerli, R., Gardner, A., West, S. A., & Griffin, A. S. (2009). Limited dispersal, budding dispersal, and cooperation: An experimental study. Evolution, 63(4), 939–949. doi:10.1111/j.1558-5646.2008.00548.x.CrossRefPubMedGoogle Scholar
  17. Kurzban, R., Burton-Chellew, M. N., & West, S. A. (2015). The evolution of altruism in humans. Annual Review of Psychology, 66(1), 575–599. doi:10.1146/annurev-psych-010814-015355.CrossRefPubMedGoogle Scholar
  18. Lay, C., Allen, M., & Kassirer, A. (1974). The responsive bystander in emergencies: Some preliminary data. Canadian Psychologist/Psychologie canadienne, 15(3), 220–227. doi:10.1037/h0081755.CrossRefGoogle Scholar
  19. Lyons, M. T. (2005). Who are the heroes? Characteristics of people who rescue others. Journal of Cultural and Evolutionary Psychology, 3(3), 245–254. doi:10.1556/JCEP.3.2005.3-4.2.CrossRefGoogle Scholar
  20. McNamee, S., & Wesolik, F. (2014). Heroic behavior of Carnegie medal heroes: Parental influence and expectations. Peace and Conflict: Journal of Peace Psychology, 20(2), 171–173. doi:10.1037/pac0000026.CrossRefGoogle Scholar
  21. Moore, D., Wigby, S., English, S., Wong, S., Szekely, T., & Harrison, F. (2013). Selflessness is sexy: Reported helping behaviour increases desirability of men and women as long-term sexual partners. BMC Evolutionary Biology, 13, 182. doi:10.1186/1471-2148-13-182.CrossRefPubMedPubMedCentralGoogle Scholar
  22. Nowak, M. A. (2012). Evolving cooperation. Journal of Theoretical Biology, 299, 1–8. doi:10.1016/j.jtbi.2012.01.014.CrossRefPubMedGoogle Scholar
  23. Nowak, M. A., & Sigmund, K. (2005). Evolution of indirect reciprocity. Nature, 437(7063), 1291–1298. doi:10.1038/nature04131.CrossRefPubMedGoogle Scholar
  24. Rand, D. G., & Epstein, Z. G. (2014). Risking your life without a second thought: Intuitive decision-making and extreme altruism. PLoS ONE, 9(10), e109687. doi:10.1371/journal.pone.0109687.CrossRefPubMedPubMedCentralGoogle Scholar
  25. Riemer, J. W. (1998). Durkheim’s “Heroic Suicide” in military combat. Armed Forces & Society, 25(1), 103–120. doi:10.1177/0095327X9802500106.CrossRefGoogle Scholar
  26. Roberts, G. (2008). Evolution of direct and indirect reciprocity. Proceedings of the Royal Society B – Biological Sciences, 275(1631), 173–179. doi:10.1098/rspb.2007.1134.CrossRefPubMedGoogle Scholar
  27. Rusch, H. (2013). Asymmetries in altruistic behavior during violent intergroup conflict. Evolutionary Psychology, 11(5), 973–993. doi:10.1177/147470491301100504.CrossRefPubMedGoogle Scholar
  28. Rusch, H. (2016). Studies on documented historical cases of civil and War heroism: A mini-review. Heroism Science, 1(1), 1–7.Google Scholar
  29. Rusch, H., & Störmer, C. (2015). An evolutionary perspective on war heroism. Militaire Spectator, 184(3), 140–150.Google Scholar
  30. Rusch, H., Leunissen, J. M., & van Vugt, M. (2015). Historical and experimental evidence of sexual selection for war heroism. Evolution and Human Behavior. doi:10.1016/j.evolhumbehav.2015.02.005.Google Scholar
  31. Smeets, P., Bauer, R., & Gneezy, U. (2015). Giving behavior of millionaires. Proceedings of the National Academy of Sciences of the United States of America, 112(34), 10641–10644. doi:10.1073/pnas.1507949112.CrossRefPubMedPubMedCentralGoogle Scholar
  32. Spence, M. (1973). Job market signaling. The Quarterly Journal of Economics, 87(3), 355. doi:10.2307/1882010.CrossRefGoogle Scholar
  33. Trivers, R. L. (1971). The evolution of reciprocal altruism. The Quarterly Review of Biology, 46(1), 35–57. doi:10.1086/406755.CrossRefGoogle Scholar
  34. West, S. A., Griffin, A. S., & Gardner, A. (2007). Social semantics: Altruism, cooperation, mutualism, strong reciprocity and group selection. Journal of Evolutionary Biology, 20(2), 415–432. doi:10.1111/j.1420-9101.2006.01258.x.CrossRefPubMedGoogle Scholar
  35. Zahavi, A. (1975). Mate selection – A selection for a handicap. Journal of Theoretical Biology, 53(1), 205–214. doi:10.1016/0022-5193(75)90111-3.CrossRefPubMedGoogle Scholar

Copyright information

© Springer International Publishing AG 2016

Authors and Affiliations

  1. 1.Public Economics GroupUniversity of MarburgMarburgGermany
  2. 2.Peter Löscher Chair of Business EthicsTU MünchenMunichGermany

Section editors and affiliations

  • Minna Lyons
    • 1
  1. 1.University of LiverpoolLiverpoolUK