Prosocial signaling and cooperation among Martu hunters

0 downloads 0 Views 453KB Size Report
Feb 19, 2015 - suggest that only prosocial generosity provides benefits through cooperation. Prosocial generosity is accompa- nied by pecuniary distancing: ...
Evolution and Human Behavior 36 (2015) 389–397

Contents lists available at ScienceDirect

Evolution and Human Behavior journal homepage: www.ehbonline.org

Original Article

Prosocial signaling and cooperation among Martu hunters Rebecca Bliege Bird ⁎, Eleanor A. Power Department of Anthropology, Stanford University, Stanford, CA, 94305

a r t i c l e

i n f o

Article history: Initial receipt 28 April 2014 Final revision received 19 February 2015 Keywords: Costly signaling Hunter–Gatherers Food sharing Trust Commitment Australian aboriginal

a b s t r a c t Among immediate-return societies, cooperative social relationships are maintained despite the lack of centralized authority, strong norms of ownership and the punishment of free-riders. The prosocial signaling theory of cooperation solves the puzzle of social cohesion in such societies by suggesting that costly forms of generosity can function as an honest signal of prosocial intent, and that the reputations for prosociality signalers build generate trust between individuals, supporting the formation of cooperative partnerships. However, not all forms of costly generosity are prosocial: we contrast two types of generosity, aggrandizing and prosocial, and suggest that only prosocial generosity provides benefits through cooperation. Prosocial generosity is accompanied by pecuniary distancing: the payment of a higher relative cost to share, and a manner of sharing that disengages the acquirer from ownership over the rights to benefit from his or her harvest. We test the prosocial sharing hypothesis among Martu hunters and find that there is a significant association between the propensity of an individual to share a higher proportion of her income and centrality in the cooperative hunting network. Those who consistently pay higher costs to share, not necessarily those who are better hunters, are preferred partners for cooperative hunting. While many have emphasized the direct, status enhancing, competitive aspects of generosity, we suggest here that prosocial generosity produces benefits indirectly, through the formation of trusting, cooperative partnerships. © 2015 Elsevier Inc. All rights reserved.

1. Introduction Under circumstances where there are no centralized institutions that ensure cooperation and promote social solidarity, how do societies remain cohesive? A substantial body of theoretical literature suggests one solution: that social cohesion is in part maintained by strong links between generosity and cooperation, generated through the honest signaling of one's intrinsic quality (Barclay, 2004, 2013; Boone, 1998; Gintis, Smith, & Bowles, 2001; Lotem, Fishman, & Stone, 2003; Macfarlan, Quinlan, & Remiker, 2013; Panchanathan & Boyd, 2004). Much of this work has focused on how seemingly costly forms of generosity, those that are performed in ways that minimize the possibility of reciprocation (such as public goods provisioning), may be an evolutionarily stable strategy under a wide range of conditions. The proposed benefits for such signaling lie in the trust that a generous individual will play fair, facilitating the formation of cooperative partnerships in other ventures. Empirical support for the prosocial signaling theory of cooperation has primarily come from experimental games, where subjects often choose to cooperate more often with those who have established themselves as more generous (Barclay & Willer, 2007; Klapwijk & Van Lange, 2009; Sylwester & Roberts, 2013; Wedekind & Milinski, 2000; Willer, 2009). Here, generosity is a strategic cost that signalers pay to influence receivers to act cooperatively toward them; the ⁎ Corresponding author. E-mail addresses: [email protected] (R. Bliege Bird), [email protected] (E.A. Power). http://dx.doi.org/10.1016/j.evolhumbehav.2015.02.003 1090-5138/© 2015 Elsevier Inc. All rights reserved.

guarantee for honesty is the potential cost to one's reputation for cheating (see Számadó, 2011). However, the design of such games has tended to make it difficult to disentangle the pathways of benefit: do generous individuals gain rewards because they are trusted cooperation partners, or because others defer to them as a result of their display of intrinsic quality? While recent work supports the formation of trust for future cooperative partnerships as the main content of the signal (Fehrler & Przepiorka, 2013), there are also cases where generosity appears to operate more as a conspicuous display of quality, meant to enhance the prestige or social dominance of the donor (Hardy, 2006; Hawkes & Bird, 2002; Mauss, 1954; Milinski, Semmann, & Krambeck, 2002; Vugt & Hardy, 2010). As others have pointed out (Sugiyama & Sugiyama, 2003), and one of us has suggested for Meriam turtle hunting (Smith, Bird, & Bird, 2003), signaling one's prowess and skill may not result in cooperative outcomes because such signaling is designed primarily to establish competitive hierarchies between individuals, a form of personal aggrandizing. As Smith et al. (2003) point out, Meriam turtle hunting carries two kinds of informative signals, sent by the ability of the hunter to pay different types of costs. Competitive (or aggrandizing) signals sent through the cost of hunting make visible the skill and ability of the hunter, particularly those who play the roles of jumper and hunt leader; more skillful individuals pay a lower cost to hunt than less skilled individuals. Prosocial signals sent through the cost of sharing the entire turtle makes visible one's political motives and prosocial motivations, mainly for hunt leaders. The material cost inherent in the way the turtle is distributed, in the hunter donating his labor to a feast-holder, in giving

390

R. Bliege Bird, E.A. Power / Evolution and Human Behavior 36 (2015) 389–397

away a whole live turtle so that all can see he took nothing for himself, all provides information to others. Men who pay such high costs to supply another's feast are indexing their support of the feast-giver; the fact that they distance themselves from distribution and from expectations of return in kind is a major component of the costly signal that allows the feast-giver to trust that the hunter values the possibility of future social interaction or cooperation more than taking an immediate individual benefit. Importantly, what the hunter is doing is visibly distancing himself from the perks of ownership, the rights to control who receives how much of the turtle, including himself. This visible distancing, so common among successful hunters in immediate-return societies, is the opposite of what we might expect to see if the display of hunter skill was entirely a form of personal aggrandizement. As such, the term competitive altruism, often used in the literature on signaling and cooperation, which includes both competitive feasting and instances of what seem to be costly sharing (Barclay, 2004), collapses two very different modes of signaling: those that are aggrandizing, and those that involve pecuniary distancing. Signaling theory suggests that key to the prosocial nature of generosity is that it is kept honest through a kind of pecuniary disinterest: a distancing of the signaler from the immediate material and immaterial profits of his or her own production (Bird & Bliege Bird, 2010; Gambetta, 2009). This distancing puts reciprocity at risk, and it is this risk that engenders trust, which provides a wide range of benefits in social interaction (Molm, Collett, & Schaefer, 2007a; Molm, Takahashi, & Peterson, 2000). Opening yourself up to potential loss in the short term is one way to let others know that you are not out for the selfish, short-term gain in social interactions. One way to accomplish this is to produce large amounts of food but conspicuously avoid profiting from that production by giving much of it away. Key to the honesty of the cost paid is that one shares a higher proportion of one's income, not simply more in an absolute sense. While aggrandizing signals often involve giving away more than others, as in competitive feasting (Boone, 1998; Dietler & Hayden, 2010; Smith & Bird, 2000) paying absolutely more than everyone else does not necessarily guarantee any prosocial intent if that cost is more easily paid. Instead, the honesty of a prosocial signal is determined by the relative cost endured: those who pay personally greater costs are those who should be trusted in cooperative partnerships. Another important component of pecuniary distancing lies in the way sharing is conducted. Building trust takes more than a single act of costly sharing: the truly prosocial sharer not only takes on the relatively higher cost of the act, she also expects no immediate payback contingent upon the provisioning of the collective good, either in deference (prestige) or in kind. Demonstrating this additional hidden quality requires cumulative signaling in ways that tell others that one's generosity does not come with hidden strings attached. Sharing in costly ways expecting to extract benefits from others contingent on the gift is a form of manipulation, not a demonstration of prosocial intent (Cronk, 1994). One of the most straightforward ways to distance one's self from the possibility of contingent sharing is to contribute to the provisioning of public goods (Smith & Bliege Bird, 2006). Other ways to subvert the manipulative power of the gift (Mauss, 1954) are to distance one's self from ownership over production by giving up all rights to distribute the goods in question, essentially making a potentially private good a collective good (Hawkes, 1993). Among many hunter gatherers, the sharing of especially large items is accomplished by someone other than the acquirer. Hunters of large animals are particularly likely to divest themselves of ownership by dropping the prey at the edge of camp, and quietly drawing no attention to their successful harvest (Hawkes, 2001). They do not participate in distribution, leaving others to determine how large a share they will receive of their own prey, and do not participate in deciding how large portions are to be that go to others (Bird & Bliege Bird, 2010). We suggest that there are at least two forms of pecuniary disinterest that ensure honesty of cooperative intent: disinterested distribution — a lack of obvious interest in contingent giving and immediate repayment (in material or immaterial

benefits), and costly sharing — taking on relatively greater (but not necessarily absolutely greater) costs to benefit others. Just as aggrandizing and prosocial signals are sent through a different set and scale of costs, so too do they result in a different set and scale of benefits. Aggrandizing signals of skill typically result in greater prestige and advancement in competitive hierarchies, which helps signalers garner attention and deference from others (Henrich & GilWhite, 2001; Macfarlan, Remiker, & Quinlan, 2012; Plourde, 2008). Prosocial signals may similarly result in heightened prestige for those who can cultivate a reputation for generosity, but the more crucial benefit may be heightened social support through the trust such signals engender. For the Meriam, it is not the competitive signals of hunter skill, but the distribution of the turtle that provides the real political benefits. As hunters themselves say, being a hunt leader that provides turtles for feasts, "ensures the old men will listen to you in meetings and support you in disputes over land, and impresses them enough to let you marry the best girl." Beyond these political benefits, empirical work suggests that while those who share more may receive benefits through bettering their position in competitive hierarchies, those who pay higher costs to share, not necessarily those who share the most, will be more likely to gain the benefits of cooperation. Gurven and colleagues (Gurven, Allen-Arave, Hill, & Hurtado, 2000) explored the relationship between generosity and the chances of receiving food when ill and found that it was those who shared a higher proportion of their production, regardless of how much of the absolute amount was shared, that were likely to be cared for when unable to produce themselves. Only those who had paid the higher relative cost in their sharing were identified as truly generous, and so received help from others confident in their cooperative partnerships. Similarly, Lyle and Smith (2014) found that contributions to public goods in an Andean village built reputations that led to greater access to cooperative partnerships, which, in turn, ultimately provided greater health-buffering benefits. Prosocial signals, then, not only build reputational standing (as aggrandizing signals do), but also appear to foster trusting, supportive relationships that have demonstrable impacts on people's livelihoods. In prosocial display, benefits do not come directly, via the generous act, but indirectly, through the benefits gained from cooperating with others. In contrast, an aggrandizing show-off would gain benefits mainly through producing more than others, sharing more but keeping more as well (Hawkes, 1991). Or, an aggrandizing producer might share only to extract obligatory repayments from others, leveraging the power of the gift to get more material benefits in return. Here, we test predictions of the prosocial signaling theory of cooperation among Australian Western Desert Martu, asking whether those who share more reap the benefits of cooperation, and whether these social benefits are gained through aggrandizing or prosocial generosity. Are aggrandizing show-offs or pecuniary distancers more likely to be rewarded for production via cooperative partnership formation? Are those who share inducing obligatory repayment or generating trust? 1.1. Predictions Our analysis focuses on cooperation and sharing in the small-prey hunting context (sandplain foot hunting). To hunt small prey on foot, Martu form dinner camps, remote temporary hunting camps several kilometers distant from the main community where the products of the day's hunt are cooked and consumed as a shared meal (see (Bird, Bliege Bird, & Codding, 2009) for details). Such hunting is dominated by women, although some men, especially older men, actively hunt small prey, bringing in about 30% of the small game calories (Bliege Bird & Bird, 2008). Individual camp members arrive at the central place either on foot or by vehicle, radiate outward from the camp to hunt or collect plant foods, and then rendezvous back at the camp at the end of the day. Dinner camps range in size from 4 to 20 people, with almost all of those present, including children older than about 5 or 6, actively working to acquire food for the shared meal. Hunters tend to leave at the same time

R. Bliege Bird, E.A. Power / Evolution and Human Behavior 36 (2015) 389–397

and time their arrival to return together just before sunset, whether they hunt cooperatively or alone. The main prey acquired on dinner camps is small game: sand monitors comprise the majority of food items, with feral cats, perentie monitors, snakes, fruit, and other small lizards making up a minority. When hunters return to camp, each hunter or team of hunters guts and cleans the sand monitors and then cooks them whole in the fire. Everyone shares the same cooking hearth, sitting around the fire drinking sweet, milky tea while the meat cooks. The sand monitors are taken out simultaneously, each hunter then receiving the cooked prey she acquired. After pulling the cooked lizards from the fire, an initial division between cooperating hunters may occur (if the hunt was cooperative). If not, each hunter simultaneously hands over individual prey to other consumers. Primary distributions between cooperating hunters are often difficult to see because hunters frequently divide the catch while still out hunting, so secondary distributions of cooked meat from hunter to consumer are the ones that are publically visible to all. Portions are nearly always handed directly to consumers (hunters and non-hunters alike), not to family groups, and if a single lizard is intended to be split between two people, say a wife and a husband, the hunter will instruct the recipient to split the prey: "This is for you and your nyupa, give him the tail," she'll say. Older children will receive portions directly, but very small children will need to be fed from their caretaker's share. During the rapid-fire tossing of almost identical lizards from one person to another in the secondary distribution, something which is incredibly difficult for a researcher to track, people may be giving with one hand while receiving with the other. Since giving and receiving is often simultaneous, a hunter has the opportunity to assess how much she is receiving and to give away more than she has actually acquired by giving away a lizard she has just received. Those who did not acquire anything at all can share by immediately redistributing all or a portion of what they have just received. The idealized outcome of secondary distribution is not for reciprocity to be balanced, but for all consumers to have the same amount to eat, and for hunters to eat very little, if any, of their own harvest, a goal which is often (but not always) met. Martu women hunt cooperatively more often than men, and we have previously shown that cooperation provides different benefits to hunters of different skill levels (Bliege Bird, Scelza, Bird, & Smith, 2012). We have previously suggested that cooperation serves as a way for poorer hunters to increase their foraging returns and for better hunters to provide social support to poorer hunters and gain the title of mirtilya, a good hunter (Bliege Bird et al., 2012). Poorer hunters want to hunt cooperatively as a form of insurance against failure and to have more to share with others, while better hunters benefit by visibly supporting poorer hunters. To gain the benefits of cooperation, partners must split the cooperative harvest evenly in the primary distribution: a hunter who captures more lizards should share those with her less successful partner. Although most harvests are evenly split, 35% are not, so there is always the potential for cooperation to fail and for each hunter to take only those lizards she captured. A hunter who claims more will be able to either keep more for herself, or to share more with others (Bird & Bliege Bird, 2010; Bliege Bird et al., 2012). The temptation to divide harvests unevenly in the initial division of the prey, allowing better hunters to gain the benefits of their overproduction, leaving poorer hunters holding the (empty) bag, suggests that (as they are not assorting on the basis of skill) hunters may determine their hunting partners on their propensity to share equally. Since primary distributions are not always visible, hunters must be keying on the widely visible secondary distributions that always take place in camp when everyone is present. We thus hypothesize that Martus use reputations for generosity built through secondary sharing with everyone at the dinner camp as a way to determine which partnerships to form for cooperative hunting: those who engage in more costly sharing (pecuniary distancers) gain the benefits of cooperation. We contrast this with an alternative hypothesis that Martu form cooperative hunting partnerships based on

391

one's foraging production: those who produce and share the most (personal aggrandizers), gain the benefits of cooperation. 1.1.1. Prediction 1 If hunters are practicing prosocial signaling, the way they share should involve distancing themselves from pecuniary reward. Honest signaling of prosocial intent requires some assurance that generosity is not intended as a form of manipulation, to extract an immediate benefit from others via deference or obligatory repayment, or that hunters are not benefitting materially from their own production by keeping more the more they acquire. Thus, we predict that those who produce more on any given day a) should not receive more from others in secondary distributions: that is, there is no immediate material in-kind exchange or intent to profit from producing more, and b) those who produce more should also share a higher percentage of the food they acquire. If, however, hunters are engaging in an aggrandizing show-off strategy, sharing may be a way to materially benefit by inducing obligatory repayment (balanced reciprocity) or to keep more for one's self. In that case, those who produce and share more should keep and/or receive more. 1.1.2. Prediction 2 If hunters are signaling prosocially, sharing in more costly ways should generate trust for cooperative partnership formation. Then, a) one's average generosity (proportion of foraging income shared) should correlate with centrality in the cooperation network, independently of total foraging production. Alternatively, if foragers are behaving more like aggrandizing show-offs, b) those who consistently produce more but do not share relatively more may have higher centrality, and production rank should correlate with centrality independently of generosity. 2. Materials and methods The analysis we report on here uses the same ethnographic data collection methodology as reported elsewhere (Bird & Bliege Bird, 2010; Bliege Bird & Bird, 2008; Bliege Bird et al., 2012). Here, we provide details of our specific predictions and the analysis of each derived dataset. Analysis of prediction 1 uses a dataset of 167 observations on 52 individual dinner-camp members that recorded the proportion of each individual's daily production that was shared to others in secondary distributions (without knowing to which particular others' prey was shared). The primary prey shared on these dinner camps are small animals, mostly sand monitor and other lizards, so there are no confounding effects of differences in prey type in this dataset. The prediction is tested using a linear mixed effects model on amount received from others (total g) per forager-day, and on percentage of total foraging income shared. Because the total amount a hunter receives from others varies across days as a function of total camp production and number in camp, we also control for this by adding it to the model as a covariate. Because there are multiple observations on the same individuals, we include individual ID as a random effect. In calculating the percentage of one's harvest shared, we use a measure of sharing that accounts for foraging income from food acquired by others as well as from the hunter's own production because hunters routinely receive up to half of their daily consumption from the efforts of others with whom they did not hunt cooperatively. Those who share a high proportion of their foraging income only when others have been more successful, and they know they will be getting some to replace what they gave are sharing a high proportion of their own production, but not a high proportion of their potential daily consumption intake. This is especially important in the dinner camp sharing context, where lizards circulate in simultaneous flows of giving and receiving and regiving. For analysis of prediction 2, we focus on 19 female and 6 male adult sand monitor hunters (out of the 52 hunters total) for whom we have

392

R. Bliege Bird, E.A. Power / Evolution and Human Behavior 36 (2015) 389–397

detailed information on hunting returns and sharing distributions following a successful sand monitor hunt. For a hunter to be included in this analysis, we had to have at least five observations of successful hunting of sand monitor lizard, and at least five observations of opportunities to share. The number of hunting observations per individual ranged from 6 to over 45, but the number of harvests on which any sharing occurred was more limited, ranging from 1 to 19. Rather than using other's perceptions of individual hunter reputations for skill and generosity, we derive rankings of skill and generosity based on our longterm observations of individual hunters. For each individual hunter we derived a ranked measure of hunting productivity based on mean foraging return rate when solitary hunting and a measure of sharing generosity (“generosity”) based on mean percentage of foraging income (meat in grams of whole carcass weight acquired and received from others) distributed per acquisition event. Foraging income includes not just the amount a hunter produces, but also what others give to her at distribution time. Generosity measures were derived from the sharing database by using a linear mixed model controlling for individual as a random effect, and total camp production as a confound. From these models, we obtained for each individual the least-squares mean percentage of all income shared and use this value as a predictor of centrality. To evaluate how foraging and sharing rank affects cooperative hunting partnerships, we observed cooperative hunting partnership formation by 52 individual adult hunters on dinner camps for small-prey hunting formed between March and August of 2004, noting which individuals shared the search and pursuit of the same prey item at least once on a hunting bout. As the population of the community at the time was close to 60 adults, we have a sufficiently comprehensive picture of cooperative hunting partnerships as they existed at that point in time. We then constructed a cooperation network (Fig. 1) consisting of a matrix of these 52 unique individuals with tie strength measured as the proportion of days cooperatively hunting when each individual was together and available to cooperate. This network has greater coverage than most social networks attempting comprehensiveness, and more than is necessary to ensure the robustness of centrality measures (Borgatti, Carley, & Krackhardt, 2006; Costenbader & Valente, 2003). The dataset includes the 25 hunters in the generosity and hunting productivity datasets. Both the cooperation and sharing data were collected during the same observation period while hunting ranks represent the long term mean hunting return rates averaged over all observation periods. We used UCINET 6.5 (Borgatti, Everett, & Freeman, 2002) to calculate the eigenvector centrality and degree centrality in the cooperative hunting network for each of our hunters in the subsample. Degree centrality measures the number of unique hunting partners for each individual; eigenvector centrality is a measure of how well connected an

Fig. 1. The cooperative hunting network. Nodes are sized by eigenvector centrality, ties are weighted according to the proportion of time individuals were together that they cooperated, and nodes are colored by consanguineal kinship ties, each color representing an extended family.

individual is to other well-connected individuals. A high centrality measure indicates that individuals tend to cooperate more often with other highly cooperative individuals. Prediction 2 is tested using a multivariate linear regression model on centrality with generosity, with hunting rank as covariate, weighted by the number of times observed. We evaluated several different transforms of both eigenvector and degree centrality to normalize their distributions and used the transform with the lowest AIC values. Eigenvector centrality is analyzed using a square root transform, while degree centrality uses a log transform. Model parameters are reported as back-transformed. Analyses were performed using JMP (v. 11, SAS Institute, 2014). 3. Results We find broad support for our predictions, which collectively suggest that Martu hunters share in ways intended to demonstrate prosocial intent, rather than to simply (or solely) demonstrate hunting prowess or to induce obligatory repayment in others. It is these prosocial signals of generosity, rather than aggrandizing signals of hunting proficiency, that leads to one being preferentially chosen by others as a hunting partner. 3.1. Do high producers benefit immediately from their overproduction? We first predicted that those who produce more should not immediately receive more from others in secondary distributions on that same day. We found that, controlling for total camp production, individual effects, and the number in camp, there is no significant relationship between how much an individual produces on that day (in grams of whole carcass weight), and how much he or she receives from others (Table 1, model 1). In fact, individuals have a tendency to receive less the more they produce (although the relationship is not significant (p = 0.0828). How much one receives is a (negative) function of how many are in camp, with people receiving 49 g less with each additional person, and a positive function of how much everyone else has produced, with hunters receiving nearly 5 g more with each additional 100 g of camp production. Secondly, there is a significant relationship between how much an individual produces and the percentage of the harvest that is shared, again controlling for total camp production, individual effects, and the number present in camp (Table 1, model 2). A 1 kg increase in harvest size increases the percentage shared by 29%, up to 86% at our maximum observed harvest size in this database of nearly 5 kg. This would suggest that the absolute amount an individual shares increases with harvest size, but that the amount an individual keeps diminishes with harvest size. Indeed, Fig. 2, which plots the raw data, shows this relationship clearly. Beyond about a 1 kg harvest (2 large or 3 average sized lizards), a successful acquirer keeps a relatively constant amount, while the amount shared increases linearly with harvest size. The fact that higher producers share a higher percentage of their harvest could be interpreted simply as marginal valuation: a larger harvest is worth less relative to its consumption value, so the costs of defending more are greater, and a greater proportion should be expected to be shared. We might be able to tease apart the effects of marginal valuation from costly sharing by comparing the average portion size kept by the best hunter on any given day with those kept by other lower-producing hunters. Given equally hungry hunters with equal resource holding potential (RHP) (likely because everyone comes back at the same time, and cooks their catch at the same time), marginal valuation would predict that portion sizes would tend to equalize, until everyone had a portion they valued equally. If everyone is equally hungry but the best hunters can claim more than everyone else, there exists some producer advantage due to unequal RHP, which would tend to weaken the argument that those who share more do not benefit materially from their overproduction. If however the best hunters claim less, this is a paradoxical result that would strengthen the argument for

R. Bliege Bird, E.A. Power / Evolution and Human Behavior 36 (2015) 389–397

393

Table 1 Regression results. Model

Outcome

Predictor

β

t

p

1

Grams of meat hunter received from others

2

Percent of daily foraging income shared

3

Amount kept by most productive hunter

4

Amount kept by the least productive hunters

5

Eigenvector centrality in cooperation

6

Degree centrality in cooperation

7

Eigenvector centrality in cooperation

8

Degree centrality in cooperation

9

Degree centrality in cooperation

10

Eigenvector centrality in cooperation

Amount acquired (g) Number in camp Total camp production (g) Amount acquired (g) Number in camp Total camp production (g) Amount acquired (g) Number in camp Total camp production (g) Amount acquired (g) Number in camp Total camp production (g) Mean generosity score Hunter rank score Mean generosity score Hunter rank score Mean generosity score Gender (M) Mean generosity score Gender (M) Mean generosity score Hunter rank score Mean generosity score Hunter rank score

−0.530 −49.000 0.046 0.028 0.418 −0.002 0.142 −23.06 0.108 0.294 −39.090 0.064 0.007 0.000 0.146 0.009 0.004 −0.011 0.037 −1.840 1.200 −0.009 0.008 −0.001

−1.83 −8.21 6.49 12.91 0.98 −5.36 1.52 −1.14 4.60 3.62 −4.52 6.39 3.12 0.16 3.06 −0.12 1.85 −2.97 1.65 −3.25 3.04 −2.07 3.83 −2.18

0.0828 0.0001 0.0001 b0.0001 0.3286 b0.0001 0.1415 .2683 0.0001 0.0006 b0.0001 b0.0001 0.0050 0.8736 0.0058 0.9077 0.0771 0.0070 0.1133 0.0036 0.0189 0.0768 0.0065 0.0652

Models 1 & 2: Linear mixed effects model with individual hunter as random effect, 167 secondary small game distributions. Models 3 & 4: Linear mixed effects models with individual hunter as random effect, 28 distributions by the most productive hunter on any given day, 72 by those who were less productive. Models 5 & 6: Linear model (square root transform for eigenvector, log transform for degree) predicting centrality in the cooperative hunting network with a hunter's average generosity in small game sharing and long term average hunting rank, 25 hunters observed over 520 total days available to cooperate. Models 7 & 8 remove hunt rank and add gender as a covariate. Models 9 & 10 focus only on the ten most frequent female hunters of small game. All p-values were reported as two-tailed.

sharing and overproduction as a signal of prosocial intent, because it is not something marginal valuation would predict. To test this, we use a linear mixed model on the amount kept with successful harvest size, controlling for number in camp (hunters and nonhunters) and total camp harvest, and adding individual ID as a random effect. We ran the model separately, first for those individuals who were the best producer on that day (n = 28 observations on 9 individuals) and secondly for all those who were not (n = 72 observations on 27 individuals). We find that for the best producers on any given day, a hunter's own harvest size had no significant effect on the amount kept (β = 0.142, p = 0.1415; Table 1 model 3). The amount acquired by others (total camp harvest) had a small but significant effect (β = 0.108, p = 0.0001), increasing amount kept by 10 g for every 100 g additional harvest size. For those who were lower producers, there was a significant, and much stronger relationship with own harvest size

Meat shared or kept (g)

4000

(β = 0.294, p = 0.0006), and an equally small effect of total camp harvest (β = 0.064, p b 0.0001) in predicting percentage shared (Table 1, model 4). Lower producers kept an additional 30 g for every additional 100 g they acquired, and kept an additional 6 g for every additional 100 g acquired by everyone else. The models predict that for any given harvest size, better producers keep 1/3 less than everyone else, and that their sharing is sensitive only to how hungry everyone else is, while the sharing of lower producers is most sensitive to how well they have done. The highest producers kept on average 296 g (±SE 54, least squares mean) while other hunters kept an average of 547 (±21). After all sharing was concluded, the best hunters ended up with significantly less than everyone else, 669 ± 76 vs. 850 ± 37. Although the sample size for daily best hunter is small and thus our failure to find an effect of own harvest size should be interpreted with caution, it does suggest that better hunters often give away so much of their own harvest that they do end up eating less than everyone else. This is not what we would expect if the more productive hunters were simply sharing according to the marginal valuation of their own harvest. 3.2. Prediction 2. Are those individuals who are consistently more generous also those who gain the benefits of cooperation? Or do the benefits flow to those who are consistently more productive?

3500 3000 2500 2000 1500 1000 500 0 0

1000

2000

3000

4000

5000

Harvest size when successful (g) Fig. 2. Plot of raw sharing data showing the effects of acquirer's harvest size (g) on the amount an individual both keeps (red) and shares (blue). Harvest size on amount shared β = 0.784, t = 26.21, p = 0.0001, r2 = .886. Harvest size on amount kept2 β = 0.0004, t = −2.28, p = 0.0249, r2 = .405.

We find that more generous hunters tend to be more central and have more cooperation partners in the cooperative hunting network. There is a significant positive relationship between a hunter's mean generosity score and both eigenvector and degree centrality (Table 1, models 5 & 6). Eigenvector centrality increases 82% with an additional 50% in average sharing generosity. However, hunt rank has no effect on either measure of centrality. We fail to support the aggrandizing prediction that centrality is a function of simply producing more than others. We find no tendency for the better hunters on average to be more central in the cooperative hunting network (Fig. 3, Table 1, models 5 & 6; Table 2 presents the raw data). For both eigenvector and degree centrality, we find that gender is a significant covariate, with men having lower centrality scores in cooperative small game hunting than women (Table 1, models 7 & 8). LS means for men's degree, controlling for sharing, are 6 times lower

394

R. Bliege Bird, E.A. Power / Evolution and Human Behavior 36 (2015) 389–397

0.7

Sqrt(Eigenvector) Leverage Residuals

0.6 0.5 0.4 0.3 0.2 0.1 0.0 -0.1 -20

-10

0

10

20

30

40

50

60

Percentage Shared Leverage, P=0.0050

70

0

5

10

15

20

25

30

HuntRank Leverage, P=0.8736

Fig. 3. Leverage plots showing the partial effects of generosity and production rank on eigenvector centrality (without controlling for gender effects).

than women's, 1.49 vs. 6.47. Men's LS eigenvector centrality independent of sharing averages 0.049; women's is 0.189. Including gender as a covariate does reduce the significance of the relationship between centrality and generosity. This seems to result primarily because a) gender and generosity covary, and b) our estimates for generosity are fraught with greater uncertainty for those women who hunt less frequently (and so are observed less frequently). If we limit our sample of female hunters to those with more than 25 hunts (n = 10) and control for any effect of hunt rank, the relationship between generosity and eigenvector centrality becomes even stronger, increasing by 0.008 for each 1% increase in generosity (Table 1, models 9 & 10). Hunt rank still has no positive effect on centrality; and in the smaller sample of women it actually has a negative (but not quite significant) effect. 4. Discussion Our results show that 1) those who acquire more small prey (sand monitor lizards) share a higher percentage of their daily income than those who acquire less — the amount (in grams) a successful hunter keeps diminishes with harvest size, while the total amount shared

Table 2 Individual centrality and generosity scores. Gender Hunt rank

Degree centrality

Eigenvector centrality

LS mean generosity

Days present

F F F M F F M F M M M F F F F F M F F M F F F F F

1 1 2 1 2 2 1 1 2 3 1 3 8 6 11 12 1 9 8 2 8 10 10 5 20

0.026 0.026 0.024 0.001 0.055 0.058 0.001 0.007 0.029 0.030 0.004 0.058 0.269 0.201 0.290 0.288 0.038 0.228 0.228 0.090 0.241 0.211 0.304 0.138 0.436

−4.14 35.89 56.02 −3.38 −0.60 59.12 −3.12 −4.14 16.99 8.70 −3.64 46.08 0.00 31.69 62.32 55.48 36.76 38.83 36.89 10.69 30.15 19.15 48.91 39.30 49.40

6 7 8 8 9 9 10 11 17 17 18 22 23 25 25 26 27 28 28 29 29 30 31 32 45

1 13 2 3 4 7 22 5 6 11 20 17 19 10 15 21 25 14 16 8 9 12 24 18 13

increases linearly with harvest size; 2) how much meat an individual receives from other hunters is not linked to how much meat he or she acquired that day; and 3) the best hunter on any given day keeps significantly less than everyone else regardless of how large his or her harvest is, a result inconsistent with simple marginal valuation (placing low value on the surplus portion of a larger harvest). We also find that 4) a hunter's average generosity with meat (percentage of all daily food income shared) predicts both eigenvector centrality and degree centrality in the small game cooperative hunting network; and 5) that a hunter's average productivity/skill (measured by his or her longterm mean foraging return rate in small game hunting) does not correlate with cooperation centrality. Finally, we show that 6) the relationship between cooperation and generosity is linked to gender, with women sharing more and cooperating more than men in small game hunting; and 7) while generosity does not predict women's cooperation centrality in the entire sample, it does for a subset of women who hunt most often. Honest signaling of prosocial intent predicts that high production does not yield a material benefit to the hunter and that generosity is not used to receive immediate benefits from manipulating others into returning favors. Because those who acquire more do not receive more from others, and do not keep more for themselves, and the best hunters in fact keep less than everyone else, better hunters do not appear to benefit from their own high production. This is peculiar indeed, because small prey like sand monitor lizards exhibit high contingency between foraging time and harvest size: hunters can thus adjust harvest size by foraging more or less than everyone else, and as we have previously shown, some hunters still produce considerably more than everyone else (Bliege Bird et al., 2012). If one is not keeping more, or receiving more from others, and in fact may be keeping less, what is the benefit of being a high producer? Our result showing that cooperation centrality is not linked to hunting productivity rank suggests that this is not among the benefits of being a high producer. Although our sample is small, and thus one should be cautious in interpreting the lack of significance as a lack of effect, the regression slope for the effects of hunk rank on eigenvector centrality is essentially zero, suggesting that if there is an effect, it is an extremely weak one. We have previously suggested that the benefits of production are gained through a kind of competitive altruism: Martu attach social status to being a good hunter, and we might assume that gaining such a reputation (and the prestige that comes with it) could be an end in itself (Bird & Bliege Bird, 2010). With the large body of literature demonstrating the reputational benefits that accrue to good hunters (Gurven & von Rueden, 2006; Kaplan & Hill, 1985a; Smith, 2004), competing in such status hierarchies seems eminently reasonable. However, the fact that cooperation centrality is not linked to one's long-term average hunting production, but is linked to one's

R. Bliege Bird, E.A. Power / Evolution and Human Behavior 36 (2015) 389–397

average generosity in sharing small game, does not support the aggrandizing implications that competitive altruism can entail. While productive/skillful hunters may be well known and may gain benefits in other arenas, this hierarchy of skill may not be the pathway through which cooperative benefits are achieved. For Martu, the path toward gaining cooperative benefits is explicit in the definition of “good hunter”, mirtilya. It is not simply to good hunters that Martu accord respect; it is to good and generous hunters, those who “hunt to share” and who share to those with whom they do not have an obligation to share (Bird & Bliege Bird, 2010; Bliege Bird & Bird, 2008; Bliege Bird et al., 2012). It seems to be this latter act that takes precedence. That better Martu hunters go to the trouble to share not just absolutely but relatively more implies that they see a heightened value to signaling their prosociality over and above signaling their skill. Our analysis here does not account for any influence of kinship on sharing or cooperation. Our results suggesting a link between cooperation and sharing could be confounded by kinship if more generous people are also more likely to share with and preferentially cooperate with kin. We think this is unlikely to be the case because previous analyses found no significant interaction between hunting productivity and cooperation with kin: all hunters, regardless of rank, were equally likely to cooperate with close and classificatory close kin when those kin were present (Bliege Bird et al., 2012). In addition, our ethnographic data suggest that more generous people are those who are less likely to cooperate and share only with kin: the definition of mirtilya (good hunter) is someone who shares not only with people they are obligated to share with (kin) but with everyone. Our analysis does suggest that gender plays a significant role in the link between cooperation and sharing. Parceling out the effects of gender on centrality is difficult because it covaries with our predictors: men tend to both share a lower proportion of their income (of small game) and to cooperate less and have lower eigenvector centrality (in small game hunting). And, with such a small sample of hunters, the differences between men and women are just visible, but the smaller range of variation among women as a group (and the small size of our sample) makes it difficult to see any significant within-group effects of sharing on cooperation. Narrowing the focus to just women who hunt more often (n = 10) does demonstrate a significant link between generosity and sharing. While the differences between men and women may be stronger, there are still subtle differences among women, and indeed, there may be substantial heterogeneity among women in the extent to which they engage in prosocial sharing. Some women seem to be less generous, and some high producers may keep more of their production for themselves. That there are such women is suggested by comments hunters make about each other: the women who score low on generosity are the same women that are often the target of complaints about stinginess and greed. These women more frequently than others will keep the largest lizard for themselves, take home more meat than everyone else at the dinner camp, or cheat a hunting partner by not sharing equally when her partner was less successful. However, even if gender differences drive much of the patterns we see, this does not affect our conclusion that sharing is linked to cooperation: there may indeed be gender and age-related differences in social strategies that influence how people share, the goals they have for sharing, and the social benefits received. We have suggested previously that women face a variety of cooperative needs that men do not face on a daily basis, and thus may be more likely to use prosocial behavior as a way to form cooperative partnerships, especially in hunting and child care (Bliege Bird & Bird, 2008; Scelza & Bliege Bird, 2008). That women in particular may use sharing as a way to discriminate between potential cooperative hunting partners is supported by the fact that women often say that they like to hunt with people who are likely to share equally with them (Scelza, 2009; Scelza & Bliege Bird, 2008). Cooperative partnerships are also important in other realms of Martu life. While we have no quantitative data on other realms of cooperation that might be facilitated by generous sharing, our substantial

395

ethnographic experience provides a few hints of how important it might be. On dinner camps, women may be able to more efficiently divide labor or trade babysitting favors. Women at different stages of their life histories focus on different resources, and cooperative specialization among women is said to be critically important for subsistence (especially in the nomadic days), as small game hunting is costlier to do while carrying small children. Women also talk about how creating strong social ties with others through generosity makes it more likely that others will hunt for you when you are ill. Cooperation is also critical for meting out traditional punishments, for elders who need to drum up support from their allies to carry out such law business. While men certainly do not seem to be using generosity in sharing of small prey to gain cooperative hunting partnerships, this does not mean that they are less generous and cooperative in general. The ethnographic data suggest that some men may be using generosity in sharing of larger prey like kangaroo and bustard (see Bird & Bliege Bird, 2010) as a way to ensure trust in the sharing of deeply secret and sacred ritual knowledge (Bliege Bird & Bird, 2008; Sackett, 1979). Women also describe their smallgame and plant food sharing as having much the same trustgenerating capacities: "It's the same for women" one woman pointed out. "If you're a generous person, you're strong in the Law." Generous women may thus find themselves accorded more trust in ritual contexts and to be able to pull off complex cooperative rituals involving actors from many communities across the Western Desert. Moreover, failure to give signals of prosocial intent and to establish cooperative partnerships may sometimes lead to exclusion from future cooperative endeavors. The process of dinner camp formation is itself a cooperative endeavor: individuals must decide with whom to form a foraging party, as the decision sets out the expectation that all in the dinner camp will have equal access to the food acquired from hunting and gathering that day. While the stated ideal is for all who express a desire to hunt to be invited along, in some cases, the 'stingier' women are sometimes conveniently overlooked in the assembly of the party that day, something easy to do without giving too much offense by simply driving past that camp last, once the hunting vehicle has already filled up with kids, ladies, and their dogs. That there is a correlation between how often women are present in our sample dinner camps, and their generosity, speaks to this: a 10% increase in generosity increases the chance a woman will be present in our recorded camps by 3% (β = .182, p = 0.0400). As Les Hiatt pointed out, in most Aboriginal societies, generosity is the main measure by which individuals are judged by others, and such evaluations matter, whether they are positive or negative (Hiatt, 1982). Rather than assuming the goal of sharing to be some vague measure of prestige or status, these data instead suggest that, at least in this case, the goal of sharing may very well be to benefit through increased access to cooperative relationships. What we had been interpreting as the benefits to high social status (gained primarily through aggrandizing signals (Bird & Bliege Bird, 2010)) should actually be more properly seen as benefits stemming from prosocial signals and the social capital they foster. Because prosocial intent is such a difficult thing to convey, it requires a multimodal signal (Higham & Hebets, 2013; Partan & Marler, 1999, 2005). Behavioral ecologists have observed that animals rarely limit their signals to one component, instead using a combination of visual, auditory, olfactory, and behavioral signals. We similarly suggest here that hunters make a series of signals, collectively aimed at conveying information about a suite of qualities that they hold. Such multicomponent signals not only increase the likelihood that the signal will be conveyed, but also improve the reliability and memorability of the signal (Rowe, 1999). When hunters spend more time to acquire more, share equally with hunting partners, distance themselves from the control of their catches, and give away a greater proportion of their daily income, this demonstrates not only that one is a skillful hunter, but also that one is contributing to the welfare of others at some cost to one's self. It is this cumulative cost, paid through a series of signals, which honestly conveys generosity and cooperative intent.

396

R. Bliege Bird, E.A. Power / Evolution and Human Behavior 36 (2015) 389–397

Prosocial sharing is not demonstrated solely by how much one shares, but it is also conveyed through the manner of sharing, how one shares: with modesty, consistency and little attention for contingency. As is common among many hunter–gatherers, a successful Martu hunter often distances himself or herself from the distribution of the harvest (see (Bird & Bliege Bird, 2010) for details). This is especially apparent when harvests consist of a prey item larger than about 1 kg in size. When a hunter practices distancing, she gives her harvest over to another non-cooperating hunter to cook and distribute, usually an older person, and has little if any influence over how the item is shared. In a sample of 43 harvests of large and small prey, 95% that were below 5000 edible kcals but only 14% above 10,000 were distributed by the acquirer. Even a harvest consisting solely of small-bodied prey items under 1 kg can engender pecuniary distancing by the hunter if a harvest includes one particularly large lizard. In that case, a hunter will usually give that one away, keeping a smaller lizard for herself. If she obtains only a single large lizard, she will usually give the entire harvest away. This is necessary in order to distinguish more prosocial sharing from other modes of sharing, including tit-for-tat reciprocity and personal aggrandizement, which are not so prosocially motivated. Singular, one-off displays of generosity are simply not sufficient to build trust; hunters must signal repeatedly and often, consistently demonstrating not only their skill, but also their generosity, modesty, and a lack of interest in any quid pro quo arrangement. Signal receivers are well aware that acts with the sheen of generosity can actually be attempts to curry favor or create debt, and so are wary of those who have not consistently and repeatedly demonstrated otherwise. In his essay on !Kung sharing, “Eating Christmas in the Kalahari” (Lee, 1969), Richard Lee describes how he tried to erase his well-established reputation for being stingy — brought about by his habit of doling out tobacco and canned food strategically and contingently — through the gift of an ox for a Christmas feast. Rather than refashioning his reputation, his one-off gift only served to create tension and suspicion about his expectations in return, and their response was to bring him down to size, rather than to give him the prestige he wanted. This negative reaction was a response to the reputation he had built for his calculus of sharing: Lee expected something in the short term for every gift, and, as he admits in the essay, gave only to manipulate others into behaving in certain ways. As Lee goes on to explain, and as others have shown as well (Wiessner, 2002), sharing large quantities of meat can be met with thanks (instead of the chastisement Lee was subjected to) and can have positive outcomes for hunters if hunters share in ways consistent with the notion that they are distancing themselves from ownership over the harvest. Wiessner recounts, from her own observations and that of other ethnographers, that when meat appears in camp, !Kung women sing and dance with elation. When meat is given away frequently and unconditionally, when the hunter keeps only a small proportion of his own production, and when the hunter practices pecuniary distancing from the kill (dropping it at the edge of camp or only casually mentioning that a very skinny, small animal is waiting to be butchered), the act of sharing is understood as a generous, otherregarding act. Through this type of sharing, others are convinced that the act is not aggrandizing or manipulative, and hunters are able to build a reputation for generosity. Wiessner suggests that through the sharing of large game, generous hunters “generate the festivities that heal rifts” and foster a collective identity of the camp as juasi kxaea khoe, “people who help each other” (2002:426). She suggests that through producing and distributing large amounts of meat, generous individuals do eventually benefit in the long term, but they benefit indirectly from creating stable, cohesive social groups that cooperate in other ways. In her discussion of reciprocity, Wiessner notes, following Hill, Kaplan, and Hawkes (1993) and Kaplan and Hill (1985b), that we must look beyond the one currency of meat to a wider set of potential returns, whether in social support, labor, other material goods, etc. We suggest here that such exchanges

are indeed essential, but should not be reduced to simple, narrow ‘reciprocation.’ Instead, we should see the bonds of support and assistance, working through many media, as the acts that build the greatest trust and solidarity and ultimately culminate in the cohesive cooperative community she suggests !Kung aspire to. This contention is supported by network analysis among Hadza, where links between cooperation and social ties of sharing also show a significant tendency for positive assortment: that is, more cooperative individuals tend to exhibit high social and spatial proximity (Apicella, Marlowe, Fowler, & Christakis, 2012). Apicella et al. hypothesize that sharing maintains these cooperative social networks, but point out that the widespread lack of control over distributions makes it unlikely that conditional reciprocity is the mechanism. Indeed, the importance of generalized reciprocity to social bonds has been demonstrated through experimental work which suggests that generalized sharing can build much stronger affective bonds of trust and commitment than contingent, negotiated exchange (Molm, Schaefer, & Collett, 2007b; Molm et al., 2000, 2007a). We suggest here a potential mechanism for ensuring the positive assortment of cooperators that does not rely on producer control and contingent sharing. 5. Conclusions Better Martu hunters do not appear to directly claim the material benefits of their overproduction. Those who produce larger harvests do not receive more from others who share to them on that day. Better hunters not only share absolutely more than poorer hunters, they share a higher proportion of what they acquire and a higher proportion of their total daily food income. Those who are consistently more generous in sharing a higher proportion of all meat they could potentially consume that day have higher centrality scores in the cooperation network. Having a higher centrality means that one is more likely to have more cooperative hunting partnerships, and that one tends to cooperate more with other highly cooperative people. Through demonstrations of generosity, both better and poorer hunters are able to signal their cooperative intent through their secondary distributions, resulting in greater access to cooperative hunting partnerships. Our results support a prosocial signaling hypothesis, providing empirical support for the theoretical links between generosity and cooperation. Generosity can be prosocial only when it is kept honest through pecuniary disinterest: a distancing of the signaler from the immediate material and immaterial profits of his or her own production, from claiming the perks of ownership. And it can only attract collaborators and strengthen social ties when it is the product of consistent, repeated interactions that allow one to build a reputation for prosociality: large one-off, showy displays of public benefit, like the Kalahari Christmas ox, are best viewed as forms of individual status competition. Unlike these obvious displays, prosocial generosity is much more likely to be subtle, and to consist of small but relatively costly acts, repeated often, and directed toward others without regard for reciprocation. Every act as complex as hunting carries with it a wide range of potential signals, but it is the prosocial signals of generosity that generate rewards through cooperation, not the aggrandizing ones. Aggrandizing signals of skill may be important in some arenas and may be more important for some foragers than others. However, the widespread phenomenon in many hunter–gatherer societies of leveling, the hunter's modesty, and the distancing of hunter from distribution, especially of larger harvests, leads us to suspect that prosocial signals will be the ones that garner more cooperative social rewards than aggrandizing ones. That most scholars (ourselves included) have given precedence to status competition is not surprising; aggrandizing and cooperative signals often co-occur because more productive hunters tend to be the most generous, so we presume the benefits they receive to be a simple extension of status competition. Drawing from the primate literature and an overly literal reading of the ideas of dominance and status hierarchies, we too easily assume that deference and prestige are the

R. Bliege Bird, E.A. Power / Evolution and Human Behavior 36 (2015) 389–397

greatest rewards, leading to humble acts of assistance by ingratiating lessers. What we see here, however, and what we think may actually hold much more widely, is rather less cut-throat and competitive. It appears to be signals of prosociality, not signals of competition, that are most sought after, and to which the rewards may be greatest. Acknowledgements The authors are grateful for the support of the Parnngurr community in conducting this research, and for helpful comments and suggestions from Douglas Bird and Brian Codding, and two anonymous reviewers. Funding for this research provided by grants from the National Science Foundation BCS-0850664, BCS-0314406, SBR-0211265, BCS-0127681, BCS-0075289; The Leakey Foundation, The Christensen Fund, The University of Maine, Stanford University Office of Teaching and Learning, The Lang Fund for Environmental Anthropology, Stanford Institute for Research in the Social Sciences, and the Woods Institute for the Environment Environmental Venture Program. References Apicella, C. L., Marlowe, F. W., Fowler, J. H., & Christakis, N. A. (2012). Social networks and cooperation in hunter-gatherers. Nature, 481, 497–501. Barclay, P. (2004). Trustworthiness and competitive altruism can also solve the “tragedy of the commons.”. Evolution and Human Behavior, 25, 209–220. Barclay, P. (2013). Strategies for cooperation in biological markets, especially for humans. Evolution and Human Behavior, 34, 164–175. Barclay, P., & Willer, R. (2007). Partner choice creates competitive altruism in humans. Proceedings of the Royal Society B: Biological Sciences, 274, 749–753. Bird, D. W., & Bliege Bird, R. B. (2010). Competing to be leaderless: Food sharing and magnanimity among Martu Aborigines. The Evolution Of Leadership: Transitions In Decision Making From Small-Scale To Middle-Range Societies. Santa Fe, NM: SAR Press, 21–49. Bird, D. W., Bliege Bird, R., & Codding, B. F. (2009). In pursuit of mobile prey: Martu hunting strategies and archaeofaunal interpretation. American Antiquity, 74, 3–29. Bliege Bird, R., & Bird, D. W. (2008). Why women hunt. Current Anthropology, 49, 655–693. Bliege Bird, R., Scelza, B., Bird, D. W., & Smith, E. A. (2012). The hierarchy of virtue: Mutualism, altruism and signaling in Martu women’s cooperative hunting. Evolution and Human Behavior, 33, 64–78. Boone, J. L. (1998). The evolution of magnanimity. Human Nature, 9, 1–21. Borgatti, S. P., Carley, K. M., & Krackhardt, D. (2006). On the robustness of centrality measures under conditions of imperfect data. Social Networks, 28, 124–136. Borgatti, S. P., Everett, M. G., & Freeman, L. C. (2002). Ucinet for Windows: Software for social network analysis. Costenbader, E., & Valente, T. W. (2003). The stability of centrality measures when networks are sampled. Social Networks, 25, 283–307. Cronk, L. (1994). The use of moralistic statements in social manipulation: A reply to Roy A. Rappaport. Zygon, 29, 351–355. Dietler, M., & Hayden, B. (2010). Feasts: Archaeological and ethnographic perspectives on food, politics, and power. Tuscaloosa: University of Alabama Press. Fehrler, S., & Przepiorka, W. (2013). Charitable giving as a signal of trustworthiness: Disentangling the signaling benefits of altruistic acts. Evolution and Human Behavior, 34, 139–145. Gambetta, D. (2009). Codes of the underworld: How criminals communicate. Princeton Univ Pr. Gintis, H., Smith, E. A., & Bowles, S. (2001). Costly signaling and cooperation. Journal of Theoretical Biology, 213, 103–119. Gurven, M., Allen-Arave, W., Hill, K., & Hurtado, M. (2000). “It’s a Wonderful Life”: Signaling generosity among the Ache of Paraguay. Evolution and Human Behavior, 21, 263–282. Gurven, M., & von Rueden, C. (2006). Hunting, social status and biological fitness. Social Biology, 53, 81–99. Hardy, C. L. (2006). Nice guys finish first: The competitive altruism hypothesis. Personality and Social Psychology Bulletin, 32, 1402–1413. Hawkes, K. (1991). Showing off: Tests of an hypothesis about men's foraging goals. Ethology and Sociobiology, 12(1), 29–54. Hawkes, K. (1993). Why hunter-gatherers work: An ancient version of the problem of public goods [and comments and reply]. Current Anthropology, 341–361. Hawkes, K. (2001). Is meat the hunter’s property? Big came, ownership, and explanations of hunting and sharing. Meat Eating and Human Evolution, 219–236. Hawkes, K., & Bird, R. B. (2002). Showing off, handicap signaling, and the evolution of men’s work. Evolutionary Anthropology, 11, 58–67.

397

Henrich, J., & Gil-White, F. J. (2001). The evolution of prestige: Freely conferred deference as a mechanism for enhancing the benefits of cultural transmission. Evolution and Human Behavior, 22, 165–196. Hiatt, L. (1982). Traditional attitudes to land resources. In R. Berndt (Ed.), Aboriginal Sites: Rites and Resource Development (pp. 47–53). Perth: University of Western Australia Press. Higham, J. P., & Hebets, E. A. (2013). An introduction to multimodal communication. Behavioral Ecology and Sociobiology, 67, 1381–1388. Hill, K., Kaplan, H., & Hawkes, K. (1993). On why male foragers hunt and share food. Current Anthropology, 34, 701–710. Kaplan, H., & Hill, K. (1985a). Hunting ability and reproductive success among male Ache foragers: Preliminary results. Current Anthropology, 131–133. Kaplan, H., & Hill, K. (1985b). Food sharing among Ache foragers: Tests of explanatory hypotheses [and comments and reply]. Current Anthropology, 26, 223–246. Klapwijk, A., & Van Lange, P. A. (2009). Promoting cooperation and trust in “noisy” situations: The power of generosity. Journal of Personality and Social Psychology, 96, 83–103. Lee, R. B. (1969). Eating Christmas in the Kalahari. Natural History, 78, 14–22. Lotem, A., Fishman, M. A., & Stone, L. (2003). From reciprocity to unconditional altruism through signalling benefits. Proceedings of the Royal Society B: Biological Sciences, 270, 199–205. Lyle, H. F., & Smith, E. A. (2014). The reputational and social network benefits of prosociality in an Andean community. Proceedings of the National Academy of Sciences, 111(13), 4820–4825. Macfarlan, S. J., Quinlan, R., & Remiker, M. (2013). Cooperative behaviour and prosocial reputation dynamics in a Dominican village. Proceedings of the Royal Society B: Biological Sciences, 280. Macfarlan, S. J., Remiker, M., & Quinlan, R. (2012). Competitive altruism explains labor exchange variation in a Dominican community. Current Anthropology, 53, 118–124. Mauss, M. (1954). The gift: Forms and functions of exchange in archaic societies. Glencoe, IL: Free Press. Milinski, M., Semmann, D., & Krambeck, H. (2002). Donors to charity gain in both indirect reciprocity and political reputation. Proceedings of the Royal Society of London. Series B: Biological Sciences, 269, 881–883. Molm, L. D., Collett, J. L., & Schaefer, D. R. (2007a). Building solidarity through generalized exchange: A theory of reciprocity. American Journal of Sociology, 113, 205–242. Molm, L. D., Schaefer, D. R., & Collett, J. L. (2007b). The value of reciprocity. Social Psychology Quarterly, 70, 199–217. Molm, L. D., Takahashi, N., & Peterson, G. (2000). Risk and trust in social exchange: An experimental test of a classical proposition. American Journal of Sociology, 1396–1427. Panchanathan, K., & Boyd, R. (2004). Indirect reciprocity can stabilize cooperation without the second-order free rider problem. Nature, 432, 499–502. Partan, S., & Marler, P. (1999). Communication goes multimodal. Science, 283, 1272–1273. Partan, S. R., & Marler, P. (2005). Issues in the classification of multimodal communication signals. The American Naturalist, 166, 231–245. Plourde, A. (2008). The origins of prestige goods as honest signals of skill and knowledge. Human Nature, 19, 374–388. Rowe, C. (1999). Receiver psychology and the evolution of multicomponent signals. Animal Behaviour, 58, 921–931. Sackett, L. (1979). The pursuit of prominence: Hunting in an Australian Aboriginal community. Anthropologica, 223–246. Scelza, B. A. (2009). The grandmaternal niche: Critical caretaking among Martu Aborigines. American Journal of Human Biology, 21, 448–454. Scelza, B., & Bliege Bird, R. (2008). Group structure and female cooperative networks in Australia’s western desert. Human Nature, 19, 231–248. Smith, E. A. (2004). Why do good hunters have higher reproductive success? Human Nature, 15, 343–364. Smith, E. A., & Bird, R. L. B. (2000). Turtle hunting and tombstone opening public generosity as costly signaling. Evolution and Human Behavior, 21, 245–261. Smith, E. A., Bird, R. B., & Bird, D. W. (2003). The benefits of costly signaling: Meriam turtle hunters. Behavioral Ecology, 14, 116–126. Smith, E. A., & Bliege Bird, R. (2006). Costly signaling and cooperative behavior. In H. Gintis, S. Bowles, R. Boyd, & E. Fehr (Eds.), Moral sentiments and material interests: The foundations of cooperation in economic life (pp. 115–148). Boston, MA: MIT Press. Sugiyama, L. S., & Sugiyama, M. S. (2003). Social roles, prestige, and health risk. Human Nature, 14, 165–190. Sylwester, K., & Roberts, G. (2013). Reputation-based partner choice is an effective alternative to indirect reciprocity in solving social dilemmas. Evolution and Human Behavior, 34, 201–206. Számadó, S. (2011). The cost of honesty and the fallacy of the handicap principle. Animal Behaviour, 81, 3–10. Vugt, M. V., & Hardy, C. L. (2010). Cooperation for reputation: Wasteful contributions as costly signals in public goods. Group Processes & Intergroup Relations, 13, 101–111. Wedekind, C., & Milinski, M. (2000). Cooperation through image scoring in humans. Science, 288, 850–852. Wiessner, P. (2002). Hunting, healing, and hxaro exchange: A long-term perspective on !Kung (Ju/’hoansi) large-game hunting. Evolution and Human Behavior, 23, 407–436. Willer, R. (2009). Groups reward individual sacrifice: The status solution to the collective action problem. American Sociological Review, 74, 23–43.