We conclude that many species possess the psychological processes to show some form of reciprocity; it might be a widespread phenomenon that varies in terms of strategies and mechanisms

Reciprocity: Different behavioural strategies, cognitive mechanisms and psychological processes. Manon K. Schweinfurth, Josep Call. Learning & Behavior, November 1 2019. https://link.springer.com/article/10.3758/s13420-019-00394-5

Abstract: Reciprocity is probably one of the most debated theories in evolutionary research. After more than 40 years of research, some scientists conclude that reciprocity is an almost uniquely human trait mainly because it is cognitively demanding. Others, however, conclude that reciprocity is widespread and of great importance to many species. Yet, it is unclear how these species reciprocate, given its apparent cognitive complexity. Therefore, our aim was to unravel the psychological processes underlying reciprocity. By bringing together findings from studies investigating different aspects of reciprocity, we show that reciprocity is a rich concept with different behavioural strategies and cognitive mechanisms that require very different psychological processes. We reviewed evidence from three textbook examples, i.e. the Norway rat, common vampire bat and brown capuchin monkey, and show that the species use different strategies and mechanisms to reciprocate. We continue by examining the psychological processes of reciprocity. We show that the cognitive load varies between different forms of reciprocity. Several factors can lower the memory demands of reciprocity such as distinctiveness of encounters, memory of details and network size. Furthermore, there are different information operation systems in place, which also vary in their cognitive load due to assessing the number of encounters and the quality and quantity of help. We conclude that many species possess the psychological processes to show some form of reciprocity. Hence, reciprocity might be a widespread phenomenon that varies in terms of strategies and mechanisms.

Keywords: Cooperation Reciprocity Cognition Emotion Norway rat Vampire bat Capuchin monkey

Introduction
The theory of natural selection predicts that only those behaviours evolve that increase the actor’s own survival and reproductive success (Darwin, 1859). Paradoxically, many species provide benefits to others, for instance, by providing care, food, information and support to con- and heterospecifics (Dugatkin, 1997). Cooperation is such a widespread phenomenon that we find evidence across the animal kingdom, ranging from bacteria (Crespi, 2001) to humans (Fehr & Fischbacher, 2003). The paradox of the evolution of cooperation was resolved for interactions between related individuals, i.e. by helping kin, shared genes are more likely to be transmitted to the next generations, which is in the interest of the helper (Hamilton, 1964). Still, the kin selection theory cannot explain the frequent occurrence of cooperation among unrelated individuals. Trivers (1971) offered a solution: reciprocal cooperation, i.e. helping those that were cooperative before. While several theoretical models have shown that cooperation can evolve via reciprocity (reviewed in Nowak, 2012), the theory has faced considerable resistance (e.g. Clutton-Brock, 2009; Connor, 2010; Hammerstein, 2003; Russell & Wright, 2009; Sánchez-Amaro & Amici, 2015; Stevens, Cushman, & Hauser, 2005; Stevens & Hauser, 2004, 2005; West, Griffin, & Gardner, 2007).
It is a central problem of the theory of reciprocity, and explaining cooperation more generally (especially among non-kin), that the extent to which behavioural exchanges are based on reciprocity is unknown. This problem is tightly linked to the cognitive underpinnings of reciprocity. Researchers, who assume reciprocity to be highly cognitively demanding, came to the conclusion that reciprocity is virtually absent in non-human animals (Amici et al., 2014; Clements & Stephens, 1995; Hauser, McAuliffe, & Blake, 2009; Pelé, Dufour, Thierry, & Call, 2009; Pelé, Thierry, Call, & Dufour, 2010; Sánchez-Amaro & Amici, 2015; Stephens, McLinn, & Stevens, 2002; Stevens et al., 2005; Stevens & Hauser, 2004). In contrast, researchers who assume that reciprocity varies in its cognitive load came to the opposite conclusion that reciprocity is widespread (Brosnan & de Waal, 2002; Brosnan, Salwiczek, & Bshary, 2010; Carter, 2014; Freidin, Carballo, & Bentosela, 2017; Melis & Semmann, 2010; Raihani & Bshary, 2011; Schino & Aureli, 2009, 2010b; Taborsky, Frommen, & Riehl, 2016). Here, we argue that the debate can be enriched and potentially resolved by identifying and discussing the concrete psychological processes of all different forms of reciprocity by using and incorporating findings from different lines of research. Although several authors have discussed the various mechanisms that may underlie reciprocal exchanges between individuals (e.g. Brosnan & de Waal, 2002; Schino & Aureli, 2010b), as far as we know, no attempt has been made to systematically relate those mechanisms to the psychological processes underlying them. Our aim in this article is to bring attention to this issue as a necessary step towards the ultimate goal of elucidating the psychological processes underlying different forms of reciprocity. Accordingly, our article is organised as follows: First, we summarise different behavioural strategies and cognitive mechanisms enabling reciprocity. Second, we apply this framework to three textbook examples of reciprocity, i.e. the Norway rat (Rattus norvegicus), the common vampire bat (Desmodus rotundus) and the brown capuchin monkey (Cebus apella). Third, we discuss the crucial behavioural, cognitive and emotional components enabling different forms of reciprocity. These psychological processes will allow us to draw clear predictions under which conditions reciprocity is likely to evolve. Finally, we provide concrete examples for prospective studies to better understand the evolutionary and psychological origins of reciprocity.

[...]

Three textbook examples of reciprocity

[...]

Norway rat
Wild Norway rats live in large multi-female, multi-male colonies, which may be composed of more than 150 individuals (Davis, 1953). They frequently interact with related and unrelated colony members and form dominance hierarchies (Calhoun, 1979). They engage in various social behaviours like alarm calls, food sharing, huddling, social grooming and social play (reviewed in Schweinfurth, under review).
Norway rats have been repeatedly shown to reciprocate help in different paradigms (reviewed in Schweinfurth in press). The most commonly used paradigm is the food-exchange setup. Here one rat, i.e. the cooperating partner, provides food via a movable platform to the focal rat (cf. Rutte & Taborsky, 2007). After a delay of up to six days, the roles are exchanged and the focal rat can provide food to its previous partner (e.g. Stieger, Schweinfurth, & Taborsky, 2017). To ensure that food donations by the focal rats are based on the previous help by a partner, focal rats are always also tested with a defecting partner that did not provide food to them. Several controls have been conducted to ensure that, for instance, differential food intake, activity, copying, or other factors cannot explain their helping levels (for a discussion, see Dolivo, Rutte, & Taborsky, 2016; Schweinfurth & Taborsky, 2018b).
Rats help each other reciprocally according to at least two behavioural strategies. First, they help each other according to direct reciprocity. Female and male rats donate food to others reciprocally by providing more food to cooperating than defecting partners (Li & Wood, 2017; Rutte & Taborsky, 2008; Schneeberger, Dietz, & Taborsky, 2012; Schweinfurth, Aeschbacher, Santi, & Taborsky, 2019; Schweinfurth & Taborsky, 2016, 2017, 2018c; Simones, 2007; Viana, Gordo, Sucena, & Moita, 2010). In addition, female rats apply direct reciprocity when grooming each other (Schweinfurth, Stieger, & Taborsky, 2017). Such reciprocal allogrooming has significant fitness benefits, as reciprocal groomers live longer and suffer less mammary tumours in the lab (Yee, Cavigelli, Delgado, & McClintock, 2008). Finally, female rats also exchange allogrooming for food and vice versa (Schweinfurth & Taborsky, 2018b; Stieger et al., 2017). Besides direct reciprocity, female, but not male, rats engage in generalised reciprocity (Rutte & Taborsky, 2007, 2008; Schweinfurth et al., 2019). In a direct comparison, however, female rats donate over 20% more food to a partner with whom they have interacted, showing that direct reciprocity generates higher levels of cooperation than generalised reciprocity (Rutte & Taborsky, 2008).
What is the mechanism underlying reciprocity in Norway rats? First, hard-wired reciprocity cannot explain reciprocity among Norway rats because they tailor their help to the partner’s helping quality (Dolivo & Taborsky, 2015) and the partner’s need (Márquez, Rennie, Costa, & Moita, 2015; Schneeberger et al., 2012; Schneeberger, Röder, & Taborsky, submitted; Schweinfurth & Taborsky, 2018a). Furthermore, rats reciprocate help by using different actions (Schweinfurth & Taborsky, 2017) and different commodities (Schweinfurth & Taborsky, 2018b; Stieger et al., 2017), making a fixed response unlikely. Second, emotion-based reciprocity is also unlikely to explain reciprocity in Norway rats. They seem not to form social bonds even after being housed together for more than one and a half years (Schweinfurth, Neuenschwander, et al., 2017). They do not accumulate social information with their partners, but rather use the last experience (Schweinfurth & Taborsky, under review; Stieger et al., 2017). Third, there is no evidence for calculated reciprocity either. The amounts of received and immediate given help are not matched in male and female rats (Schweinfurth et al., 2019; Schweinfurth & Taborsky, under review). This is in line with their numerical ability being limited to six items, which is below the amount of grooming bouts they are known to reciprocate (cf. Davis & Bradford, 1986; Schweinfurth, Stieger, et al., 2017).
Reciprocity in Norway rats is probably best explained by attitudinal reciprocity (reviewed in Schweinfurth, in press). Rats form attitudes that are based on the last encounter with a partner (Schweinfurth & Taborsky, under review). Importantly, using the last encounters for reciprocity is not a result of memory interference as rats have been shown to memorise several partners (Kettler, Schweinfurth, & Taborsky, under review), several food preferences (Galef, Lee, & Whiskin, 2005) and several unique events (Panoz-Brown et al., 2016) despite a long time delay with potentially disruptive experiences. Importantly, attitudes are linked to received cooperation and not just the product of ‘feeling good’, as a study showed in which rats refused to reciprocate food donations that they received in the presence of, but not by, another rat (Schmid, Schneeberger, & Taborsky, 2017). Attitudes can be generalised to other partners because female rats show not only direct but also generalised reciprocity (Rutte & Taborsky, 2008). In addition, attitudes are not binary responses, like a cooperator or defector tag, but can be modulated by different values of help. For instance, rats groom a partner more often that has provided food to them than vice versa, suggesting that one food item is valued more than one allogrooming bout (Schweinfurth & Taborsky, 2018b). In addition, rats are more likely to provide oat flakes to a partner that provided them with banana pieces, i.e. highly preferred food, than with carrot pieces, i.e. less preferred food (Dolivo & Taborsky, 2015).


Common vampire bat
Vampire bats, which include three species, feed exclusively on the blood of other mammals (Dalquest, 1955). Common vampire bats live in small groups of eight to twelve individuals (Wilkinson, 1984). Such groups roost together in large colonies, ranging from a few individuals to over 2,000, whereby females often move between different roosts (Wilkinson, 1988). Groups usually consist of one male and its female harem or male bachelor groups (Wilkinson, 1985b, 1985a). Vampire bats are the only bats that regurgitate food to donate it to others (Carter & Wilkinson, 2013a). Besides food donations, vampire bats show high levels of allogrooming compared to other bats (Carter & Leffer, 2015). In addition, they adopt and nurse offspring of other colony members (Carter & Wilkinson, 2013a; Wilkinson, Carter, Bohn, & Adams, 2016).
Common vampire bats regurgitate blood to donate it to others. Early studies used numerous observations of these bats in their natural habitat and found that given and received help is correlated and reciprocated (Denault & McFarlane, 1995; Wilkinson, 1984). Recently, these findings have been replicated (Carter & Wilkinson, 2013c) and extended by several controlled experiments with captive bats (reviewed in Carter & Wilkinson, 2013b). In several experiments, individual focal bats were removed from their colony and fasted for one day. After the hungry focal individual was returned to its group, all food donations to this individual were recorded. The focal individuals received blood mostly from partners, which they had provided food before (Carter & Wilkinson, 2013c).
They mainly donate blood with conspecifics with which they roost together frequently but that do not necessarily belong to their group (Carter & Wilkinson, 2013b). Food donations are more common between females, which form stable social bonds, but males have also been shown to regurgitate blood for others in the laboratory (Carter & Wilkinson, 2013b). Such donations are highly valuable to recipients because vampire bats die within two to three days without a blood meal (Freitas, Welker, Millan, & Pinheiro, 2003; McNab, 1973). They reciprocate blood with both kin and non-kin, whereby donations are better explained by reciprocity than by relatedness (Carter & Wilkinson, 2013c). Extending the network to non-kin has been shown to be beneficial for the bats because when their main association partner was temporarily removed, those that had more associations with unrelated roost mates received more food donations (Carter, Farine, & Wilkinson, 2017; Carter & Wilkinson, 2015b). Besides exchanging blood, the bats also exchange allogrooming for food according to direct reciprocity (Carter & Wilkinson, 2013c; Wilkinson, 1986). In contrast, they seem not to use generalised reciprocity (Carter & Wilkinson, 2013c).
What is the mechanism underlying reciprocity in common vampire bats? First, hard-wired reciprocity cannot explain reciprocity since the bats exchange different services, like allogrooming with food (Carter & Wilkinson, 2013c). In addition, reciprocity is limited to few closely bonded individuals (Carter et al., 2017), which makes automatic responses to received help unlikely. Second, attitudes based on recent encounters seem to be unable to generate reciprocity because reciprocity can only be detected over long time frames or when socially bonded bats are observed in captivity (reviewed in Carter & Wilkinson, 2013b; Wilkinson et al., 2016). Third, calculated reciprocity is an unlikely explanation for their reciprocity. Like Norway rats, common vampire bats do not match the actual amount of received and given help (Carter & Wilkinson, 2013c), which makes reciprocity based on calculations unlikely. However, little is known about their cognitive capabilities that could elucidate the underlying cognitive processes that they use (for the only studies of examining cognitive skills, see Ratcliffe, Fenton, & Galef, 2003; Vrtilek, Carter, Patriquin, Page, & Ratcliffe, 2018).
Reciprocity in vampire bats is probably best explained by emotion-based reciprocity. Rather than considering the last encounters as Norway rats do, the bats reciprocate help over long time spans (see above). They form enduring social bonds with kin and non-kin, which can last over more than ten years (Carter et al., 2017; Carter & Wilkinson, 2013c, 2015b; Wilkinson, 1985b, 1985a, 1986). Importantly, their food-sharing network closely mirrors their social-bonding network (Carter & Wilkinson, 2013c). This suggests that social bonds play a crucial role in their decisions to help conspecifics. In line with this, the amounts of donated blood and grooming are linked to oxytocin levels, which suggests an emotional component in the decision to reciprocate (Carter & Wilkinson, 2015a).


Brown capuchin monkey
Brown capuchin monkeys, also called tufted or black-capped capuchin monkeys, live in colonies of up to 30 individuals (Carosi, Linn, & Visalberghi, 2005). They show a distinct linear hierarchy in both sexes (Janson, 1985). Further, their breeding system can be described as either one-male, multi-female or multi-male, multi-female (Carosi et al., 2005). These monkeys show various social behaviours, such as alarm calls, collaborative hunting, food sharing, social grooming and social play (Fragaszy, Visalberghi, & Fegan, 2009; Izawa, 1980).
Brown capuchin monkeys are probably the best-known example for reciprocity. Reciprocity has been demonstrated repeatedly by using various methods under controlled captive conditions. For instance, they donate food to others by handing over or dropping food close to a partner in an adjacent compartment, who will return the favour in the same way (de Waal, 2000; see de Waal, 1997, for a detailed ethogram of the donations). To ensure that food donations are a product of received help and partner directed, several control conditions have been conducted with partners of differing relationship quality or partners being absent (reviewed in de Waal & Brosnan, 2006). In addition to these active and passive food transfers, the monkeys also reciprocate food donations by using various less intuitive testing apparatuses, i.e. bar pulling, joysticks, lever boxes and token exchanges (Hattori, Kuroshima, & Fujita, 2005; Mendres & de Waal, 2000; Parrish, Brosnan, & Beran, 2015; Suchak & de Waal, 2012).
All the tests described above investigated direct reciprocity. In addition, capuchin monkeys show generalised reciprocity (Leimgruber et al., 2014). As far as we are aware, there has been no published report of indirect reciprocity in brown capuchin monkeys. However, they pay attention to third-party interactions and are inclined to accept exchange offers from humans that were observed to frequently reject such exchange requests from other monkeys (Anderson, Kuroshima, Takimoto, & Fujita, 2013). This might suggest indirect reciprocity, but further studies are needed that directly test this possibility.
What is the mechanism underlying reciprocity in brown capuchin monkeys? In contrast to Norway rats and common vampire bats, the mechanism of reciprocity among these monkeys is less obvious. The only mechanism that can be almost certainly excluded is hard-wired reciprocity. When the monkeys help each other, they consider the quality of received help (de Waal, 2000) and the quality of their relationship (Sabbatini, De Bortoli Vizioli, Visalberghi, & Schino, 2012). This makes a fixed response unlikely. Furthermore, calculated reciprocity seems unlikely but possible because two studies found that the amount of received and given help was correlated, which suggests memory and score keeping of previous helpful events to some degree (e.g. de Waal, 1997, 2000). In fact, capuchin monkeys have been shown to add items to a pool of non-visible items, suggesting that they can keep track of multiple food donations, which is needed for calculations (Beran, Evans, Leidghty, Harris, & Rice, 2008). However, not all studies found such a correlation between received and given help (Brosnan, Freeman, & de Waal, 2006; Sabbatini et al., 2012; Suchak & de Waal, 2012). Finally, studies set out to test for calculated reciprocity found no evidence (Amici et al., 2014; Pelé et al., 2010). It should be noted, however, that the monkeys in these studies did not pass the task-understanding condition. Therefore, additional studies on calculated reciprocity are needed.
There is less ambiguous evidence for the two remaining mechanisms underlying reciprocity. There is good evidence for attitudinal reciprocity (reviewed in Brosnan & de Waal, 2002; de Waal & Brosnan, 2006). In contrast to common vampire bats and in accordance with Norway rats, brown capuchin monkeys show short-term reciprocity (see above). Furthermore, they reciprocate help with familiar and unfamiliar partners (Suchak & de Waal, 2012), which suggests short-term attitudes rather than long-term bonds being the reason to help. However, there is also good evidence for emotion-based reciprocity. In a direct comparison, the monkeys were more likely to allow access to their food to a socially bonded individual than to an individual that provided food to them before (Sabbatini et al., 2012). This study suggests that the monkeys probably cooperate over long time frames and an emotional bond can become more important that attitudinal reciprocity in a partner choice situation. This is in line with observations from the wild that showed evidence for long-term reciprocity (di Bitetti, 1997; Izawa, 1980; Schino, di Giuseppe, & Visalberghi, 2009a; Schino, Di Giuseppe, & Visalberghi, 2009b; Tiddi, Aureli, Polizzi di Sorrentino, Janson, & Schino, 2011; Tiddi, Aureli, & Schino, 2012).
Hence, reciprocity in brown capuchin monkeys is probably a result of attitudinal reciprocity and emotion-based reciprocity. While attitudinal reciprocity cannot explain the selective helping of bonded individuals, emotion-based reciprocity cannot explain the repeatedly observed short-term reciprocity. Still, the results must not be contradictory, and both may explain some aspects of cooperation in these monkeys. For instance, Sabbatini et al. (2012) found that the same monkeys show short-term reciprocity in dyads, but long-term reciprocity in trios. This is an interesting finding because it suggests that individuals can possess multiple mechanisms to achieve reciprocity. The same may apply to other species, although this has not been studied extensively other than in humans.
Humans use different mechanisms to achieve different forms of reciprocity. People employ a calculated reciprocity approach with unfamiliar (Andreoni & Miller, 1993; Gachter & Falk, 2002) or business partners (Anderson, Hakansson, & Johanson, 1994; Steidlmeier, 1999). If, however, it gets difficult to memorise several events with a partner, people focus on the attitude towards partners based on the last encounter (Milinski & Wedekind, 1998). In contrast, we rarely use short-term reciprocity when interacting with friends; instead, we rely on emotional bonds based on long-term reciprocity (reviewed in Massen, Sterck, & de Vos, 2010; Silk, 2003). Overall, we help friends more than strangers (Gächter, Kessler, & Königstein, 2011) because we trust them (Buchan, Croson, & Dawes, 2002; Majolo et al., 2013) and expect them to return favours (Deutsch, 1975; Walker, 1995). Interestingly, donations towards strangers can be increased by applying oxytocin, which is associated with emotional bonds (Zak, Stanton, & Ahmadi, 2007). In addition, the more often strangers interact, the more trust is built up and the more they are treated like friends (Gächter et al., 2011). This suggests that we use multiple mechanisms that can transform into each other rather than being static.