International Journal of Psychology, 2015

DOI: 10.1002/ijop. 12215

Direct reciprocity in animals: The roles of bonding and

affective processes
Esteban Freidin1,2 , Fabricio Carballo1,3 , and Mariana Bentosela1
1
Grupo de Investigación del Comportamiento en Cánidos (ICOC), Instituto de Investigaciones
Médicas “Alfredo Lanari”, CONICET/UBA, Buenos Aires, Argentina
2 Instituto de Investigaciones Económicas y Sociales del Sur (IIESS), CONICET, Bahía Blanca,

Argentina
3 Instituto de Investigaciones Biológicas y Biomédicas del Sur (INBIOSUR), CONICET/UNS,

Bahía Blanca, Argentina

T he presence of direct reciprocity in animals is a debated topic, because, despite its evolutionary plausibility, it is
believed to be uncommon. Some authors claim that stable reciprocal exchanges require sophisticated cognition which
has acted as a constraint on its evolution across species. In contrast, a more recent trend of research has focused on the
possibility that direct reciprocity occurs within long-term bonds and relies on simple as well as more complex affective
mechanisms such as emotional book-keeping, rudimentary and higher forms of empathy, and inequity aversion, among
others. First, we present evidence supporting the occurrence of long-term reciprocity in the context of existing bonds
in social birds and mammals. Second, we discuss the evidence for affective responses which, modulated by bonding,
may underlie altruistic behaviours in different species. We conclude that the mechanisms that may underlie reciprocal
exchanges are diverse, and that some act in interaction with bonding processes. From simple associative learning in social
contexts, through emotional contagion and behavioural mimicry, to empathy and a sense of fairness, widespread and
diverse social affective mechanisms may explain why direct reciprocity may not be a rare phenomenon among social
vertebrates.

Keywords: Prosociality; Direct reciprocity; Empathy; Fairness; Inequity aversion; Bonding; Oxytocin.

Direct reciprocity involves the sequential perfor- delay between delivering a benefit to another agent
mance of costly behaviours that produce benefits to and this agent’s reciprocation opens up the door to
the interacting individuals (we here do not deal with the possibility of the latter taking the benefit without
negative—vengeful—reciprocity; Carter, 2014). This reciprocating the favour received (Hauser, McAuliffe, &
phenomenon can be distinguished from others such as Blake, 2009). Despite these difficulties, some conditions
indirect and generalised reciprocity: direct reciprocity may facilitate the stability of reciprocal exchanges. For
is involved when A acts beneficially towards B after instance, a prospect for repeated encounters could make
B benefited A; indirect reciprocity when A benefits B nonreciprocation a less attractive alternative. This is so
after knowing that B acted beneficially towards C; and because mutual cooperation may provide a higher payoff
generalised reciprocity when A benefited B after C bene- when accumulated in the long run (Seyfarth & Cheney,
fited A (Carter, 2014). In this article, we focus on direct 2012). In addition, the ability to discriminate cooperative
reciprocity only, and we discuss some of the proximal types may allow reciprocal agents to direct their altruism
mechanisms that may underlie it. in a conditional manner, thus protecting themselves from
The obstacle for direct reciprocity to become evo- exploitation by noncooperators (Yamamoto & Takimoto,
lutionarily stable is that altruistic behaviours (costly 2012).
voluntary acts that benefit others—by “costs” we mean Direct reciprocity is generally accepted to be com-
short-term fitness costs, not life-time net fitness costs) mon among people (Brosnan & de Waal, 2014; Carter,
risk being exploited by cheaters or noncooperators. The 2014). In contrast, many authors maintain that direct

Correspondence should be addressed to Esteban Freidin, Instituto de Investigaciones Económicas y Sociales del Sur (IIESS), CONICET Bahía
Blanca, Bahía Blanca, Argentina. (E-mail: efreidin@iiess-conicet.gob.ar).

© 2015 International Union of Psychological Science

2 FREIDIN, CARBALLO, BENTOSELA

reciprocity is a rare phenomenon in the animal king- preferences, such as empathic concern and a sense of fair-
dom (e.g. Clutton-Brock, 2009; Hauser et al., 2009). On ness (Yamamoto & Takimoto, 2012).
one hand, reports of direct reciprocity in non-human ani- In short, affective mechanisms may contribute to the
mals have been usually challenged on grounds of alter- stability of direct reciprocity by helping to build partner-
native explanations, such as altruism directed towards ships and social bonds that protect altruistic tendencies
kin, interactions that produce immediate mutual bene- from exploitation by selfish individuals.
fits or situations that involve coercion, among others (see In this article, we focus on the case of animal reci-
Clutton-Brock, 2009). Associated with this scepticism for procity in the context of affective bonds and relationships
animal reciprocity, it is the view that direct reciprocity beyond the short term. We aim at presenting current
relies on sophisticated cognitive abilities. Some authors debates on the comparative psychology of the affective
claim that the delay between the altruistic act and its mechanisms of direct reciprocity and we analyse the
reciprocation requires an ability to recognise partners’ evidence for the modulation of these processes within
identity, skill with numbers to maintain strict counting of bonds. The present discussion of relevant literature does
favours given and received, as well as, inhibitory control not attempt to be exhaustive. For limits of space and refer-
to restrain oneself from exploiting others and spoil future ences, we count on selected examples with a focus on the
cooperation (Hauser et al., 2009). Even more, providing latest influential research, and also rely on other authors’
benefits conditionally on others’ past behaviour may need reviews of related topics (and thus do not cite seminal
an adequate memory to track others’ reputation and an papers if not strictly needed). Our attempt is to clarify the
ability to update it based on direct and indirect interac- reach of the empirical studies cited, and to show and sug-
tions (Hauser et al., 2009). On this view, direct reciprocity gest promising avenues for future research on the topic
is seen as a deeply calculated affair, and, some authors of the affective processes underlying direct reciprocity.
think that such demanding calculations could have lim- The rest of the manuscript is structured as follows. In
ited its presence across species (e.g. Clutton-Brock, 2009; the next section, called Reciprocity in Long-term Bonds,
Hauser et al., 2009). In this sense, a particular mechanistic we mention some of the evidence for reciprocity based on
hypothesis about direct reciprocity (i.e. calculated reci- stable bonds in different species, as a nonexclusive alter-
procity) has reinforced the view that animal reciprocity native to the hypothesis of calculated reciprocity which
may be a rare phenomenon. dominated the literature in the last decade. In the fol-
The scepticism about animal reciprocity has, nonethe- lowing section, called Affective Processes Modulated by
less, not been homogeneously held in the academic com- Bonding, we analyse and critically discuss the evidence
munity (e.g. see Schino & Aureli, 2009 and Carter, 2014 for different affective responses which may help regu-
for views that contrast with those of Clutton-Brock, 2009 lating altruism in the context of long-term relationships.
and Hauser et al., 2009). Many authors have posited Finally, we close with a section in which we conclude
alternative mechanisms through which direct reciprocity that an array of diverse mechanisms, some simple and
could be sustained, and have claimed to find instances of widely present across species, others more complex and
direct reciprocity in birds and mammals, going from rats thus not as pervasive, may contribute to sustaining direct
through monkeys to apes (see Carter, 2014 for a review). reciprocity in animals. In particular, we discuss the poten-
Indeed, many findings provide support, not so much for tial role of associative learning processes in bonding and
calculated reciprocity, but instead for a more emotionally reciprocity.
based type of altruism (Carter, 2014; Schino & Aureli,
2009). RECIPROCITY IN LONG-TERM BONDS
Many social animals establish long-term bonds within
which reciprocal exchanges seemingly occur (e.g. see Among the requirements for demonstrations of direct
Schino & Aureli, 2009; Seyfarth & Cheney, 2012). In con- reciprocity, the contingent exchange of altruistic
trast to the idea of cognitive constraints on reciprocity, behaviours between reciprocators has been at the centre
evolution could have shaped animals’ motivations and of the debate (see Clutton-Brock, 2009 and Carter, 2014
affective responses, inclining them to engage in long-term for different views). On one perspective, proof of direct
reciprocal interactions. In this sense and contrary to the reciprocity demands a short-term association between
view of reciprocity as a calculated affair, direct reci- reciprocators’ altruism. That is the logic underlying
procity may not necessarily require sophisticated cogni- Tit-for-Tat (Carter, 2014). In this sense, as proof of
tion but may rely on affective mechanisms that help to reciprocity, some authors expect to find a short-term
establish and regulate bonding (Schino & Aureli, 2009). causal chain between an animal’s altruistic behaviour
These motivational processes could range in complex- and its partner reciprocation (e.g. Clutton-Brock, 2009;
ity from simple associative learning in social contexts, Hauser et al., 2009). On a contrasting view, other authors
through reflexive emotional responses, such as emotional pose that favours may balance out on the long run (e.g.
contagion and behavioural mimicry, to other-regarding Carter, 2014), which is particularly sensible for long-term

© 2015 International Union of Psychological Science

 DIRECT RECIPROCITY IN ANIMALS 3

relationships such as what we call friendship in humans In synthesis, the idea that direct reciprocity may occur
(Seyfarth & Cheney, 2012). Indeed, long-term bonds and be stabilised in the context of long-term bonds has
satisfy the conditions that facilitate the stability of direct attracted attention in recent years, has received empiri-
reciprocity, such as repeated encounters and conditional cal support in different species, and has been established
cooperation (Schino & Aureli, 2009). as a complementary, presumably, more general (than cal-
The view of direct reciprocity as a rare cognitively culated reciprocity) basis for reciprocity in the animal
demanding phenomenon was dominant in the last decade, kingdom. These affiliative processes may in turn modu-
but has receded after the recognition of animals’ altruistic late other socio-cognitive mechanisms such as different
exchanges under field conditions and new evidence for forms of empathy and perceptions of fairness (Yamamoto
alternative hypotheses about its underlying mechanisms. & Takimoto, 2012), which may also play a role in recip-
Along this line, reciprocal exchanges have been postu- rocal exchanges, as we discuss in the next section.
lated to be based on affective mechanisms more than in
a calculated assessment of favours given and received
AFFECTIVE PROCESSES MODULATED BY
(see Schino & Aureli, 2009, for a review of long-term
BONDING
reciprocal exchanges in primates). Several examples
illustrate this.
In this section, we first discuss recent research on
For instance, Sabbatini, De Bortoli Vizioli,
emotional contagion, vicarious learning, behavioural
Visalbergui, and Schino (2012) evaluated capuchin mon-
mimicry and empathic concern, as mechanisms that
keys (Cebus apella) in triads and found that subjects chose
could underlie altruism within social bonds (see Cronin,
who to interact with based on existing long-term bonds
2012, for a recent analysis of the effect of bonding on
more than on recent food transfers among them. Schino
prosociality—a concept related to altruistic behaviour,
and Aureli (2009) reviewed other studies of direct reci-
but without assumptions about costs-, especially in
procity within long-term bonds in primates. In the case
non-human primates). And second, we end with a crit-
of non-primates, there are suggestive cases of reciprocity
ical discussion of experimental research on inequity
in vampire bats and different bird species (Carter, 2014).
aversion (IA) as a mechanism presumed to aid partner
The case of bats (Desmodus rotundus) is interesting
selection and contribute to the stability of bonds. Other
because they commonly share meals in nature. Success-
relevant processes that may be modulated by bonds,
ful foragers regurgitate blood to feed individuals that were
less successful on that day (Carter & Wilkinson, 2013). such as, for example, communication and consolation
This behaviour was also studied in controlled conditions behaviours, are not discussed in this review for a matter
in captivity. Carter and Wilkinson (2013) found that of space.
donors approached a fasted bat, and that the food donated
was better explained by the history of prior exchanges From emotional contagion through vicarious
between them than by kinship. Moreover, sharing dyads learning to empathic concern
showed allogrooming patterns that were consistent
across long periods of time, suggesting an important role The concept of empathy has been associated with a
of bonding in their reciprocal altruistic behaviours. broad spectrum of psychological phenomena, from the
The effect of long-term bonding on reciprocal experience of emotions that match and are triggered by
exchanges of different currencies has been studied another’s emotions, going through recognition of other’s
not only for appetitive behaviours, such as food sharing thinking and feeling, to empathic concern which may
or grooming, but also for behaviours related to agonistic induce helping behaviours (see Panksepp & Panksepp,
support. In a field study with macaques, Macaca syl- 2013 for a review about cross-species research on
vanus, the formation of coalitions in the reproductive empathy).
season could be predicted by their bonds weeks before, It has been hypothesised that rudimentary and more
suggesting that these primates cooperate in risky sit- complex forms of empathy evolved in the context of
uations based on previously established relationships parental care and pair bonding and later expanded to affect
(Berghänel, Ostner, Schröder, & Schülke, 2011). Similar other relationships. Indeed, authors argue that advanced
results have been found in a study with ravens (Corvus forms of empathy are preceded by and grow out of
corax) in captivity (Fraser & Bugnyar, 2012). The authors more elementary ones (Panksepp & Panksepp, 2013).
recorded episodes of agonistic support and found that, From a proximal perspective, the affective attunement
besides kinship and dominance, ravens were more likely of individuals may aid the recognition of others’ needs,
to support those that had provided agonistic support to which, under certain circumstances, may incline individ-
them and preened them within the previous week. In uals towards altruism (e.g. see Schneeberger, Dietz, &
this sense, Fraser and Bugnyar (2012) concluded that Taborsky, 2012). This mechanism on its own would, how-
agonistic support was dependent on reciprocation within ever, not be evolutionarily stable because nonreciprocat-
bonds more than in a short-term tit-for-tat strategy. ing unempathetic individuals would be able to exploit it

© 2015 International Union of Psychological Science

4 FREIDIN, CARBALLO, BENTOSELA

in others. Accordingly, emotional contagion and differ- Empathic concern

ent forms of empathy may be modulated by the bonding
relationship between individuals, suggesting that animals Empathy can be understood as the ability to recognise
are not unconditionally altruistic. and share others’ feelings (Panksepp & Panksepp, 2013;
Singer, 2008). Demonstrations of empathy in non-human
animals are debated and controversial; therefore, its exis-
Emotional contagion and behavioural mimicry tence beyond humans is not agreed (see Edgar, Nicol,
Clark, & Paul, 2012 for a review and discussion of
Emotional contagion is a psychological phenomenon methodological issues associated with evaluating empa-
in which the perception of an affective behavioural change thy in nonverbal animals). Nonetheless, some authors
automatically activates the same process in another indi- believe that empathy for other’s distress could be a source
vidual (Panksepp & Panksepp, 2013). The most common of altruistic actions (e.g. Yamamoto & Takimoto, 2012)
example of emotional contagion in humans is infectious and have attempted to show that in animals.
crying among babies (Homo sapiens), whereas yawning Bartal, Decety, and Mason (2011) investigated whether
represents another paradigmatic contagious behaviour rats showed any concern for others’ welfare. In their
(see Palagi, Norscia, & Demuru, 2014). Indeed, conta- study, pairs of rats that lived in the same cage for a
gious yawning has been shown in primates, including few weeks before the experiment (intended to enhance
bonobos (Pan paniscus) and humans, and has been found their familiarity) were then tested in a rescue task. One
to occur more frequently between individuals with a close of the rats (the trapped rat) was placed in a restraining
social bond (e.g. Palagi et al., 2014), which is consistent tube, and its cage mate (the free rat) could apply a force
with the idea of the social modulation of empathy-related to the restraining apparatus to open it. Indeed, free rats
abilities. opened the door more frequently in the experimental than
A phenomenon related to behavioural contagion, in control conditions.
behavioural mimicry, not only occurs within pre-existing The authors claimed that these findings provided evi-
bonds but may also modulate these bonds. For instance, dence for a behaviour motivated by an empathic concern
capuchins monkeys were exposed to two persons, one (Bartal et al., 2011). Nevertheless, this interpretation has
who imitated them and the other who performed con- been criticised on different fronts. First, similar rescue
tingent actions without imitation (Paukner, Suomi, behaviour in ants (Cataglyphis cursor) suggests that
Visalberghi, & Ferrari, 2009). The authors found that other explanations may need to be discarded first before
several of the monkeys’ behaviours (looking, approach- confidently referring to rats’ behaviour as empathic
ing, exchanging tokens and spending time in proximity) (Vasconcelos, Hollis, Nowbahari, & Kacelnik, 2012).
were preferentially directed towards the imitator and/or For example, rats’ liberating behaviour could be simply
away from the non-imitator, suggesting a change in explained by the reinforcing effect of terminating stress
the monkeys’ preferences after being exposed to them. signals from the trapped companion (Edgar et al., 2012;
Similar phenomena have been reported in humans. For Panksepp & Panksepp, 2013). Second, Bartal et al.’s
instance, nonconscious mimicry in adults increased (2011) findings could involve post-rescue social rein-
prosociality from the mimicked person (van Baaren, forcement associated with social contact. In fact, Silber-
Holland, Kawakami, & van Knippenberg, 2004). berg et al. (2014) could not replicate Bartal et al.’s find-
Vicarious learning may rely on contagious experiences ings when they controlled for this. In their study, when the
as well (Panksepp & Panksepp, 2013). Many researchers rats had not learned how to free their cage mates, they did
studied this phenomenon using aversive conditioning in not acquire this behaviour in a condition in which social
rodents. In this type of protocol, an observer watches a contact was not possible (Silberberg et al., 2014). This
target individual receive a shock after the presentation last result suggests that a motivation for social contact
of a conditioned stimulus (CS) or in a specific context. may have been the main drive underlying rats’ altruism.
Then, when tested in the presence of the CS or in the cor- Despite the inconclusive results of Bartal et al.’s
responding context, the observer typically shows learned study, other evidence points in the direction for a role
fear responses, despite having never experienced shocks of empathy-related abilities in reciprocity. Schneeberger
in those circumstances. In agreement with results from et al. (2012) investigated whether reciprocal cooperation
contagious and mimicked behaviour, empathetic pain can between unrelated Norway rats (Rattus norvegicus) could
be modulated by the pre-existing relationship between be modulated by the relative need of the recipient. These
social partners (see Panksepp & Panksepp, 2013). This authors systematically varied the hunger status of the
has led researchers to wonder whether emotional con- recipient rat in a generalised reciprocity paradigm. In
tagion, vicarious learning and other potential empathic their main treatment, the focal rat received food from an
experiences may lead animals to act altruistically (e.g. unknown trained rat, and then she had to choose whether
Chang, Barter, Becket Ebitz, Watson, & Platt, 2012), as to pull a tray to provide food for another rat (the recipi-
we discuss in the next subsection. ent). Among other results, Schneeberger et al. found that

© 2015 International Union of Psychological Science

 DIRECT RECIPROCITY IN ANIMALS 5

focal rats gave more food to lighter than heavier receivers the fact that there are no empirical studies that directly
in the food deprived condition, showing that rats were relate IA with bond-based reciprocity in non-humans,
able to recognise the need state of social partners and to preliminary observations suggest that IA might be
accommodate their altruistic helping accordingly. These modulated by bonding.
results do not directly speak of the effect of empathy on In the rest of this subsection, we focused on discussing
direct reciprocity, but show that rats’ altruism was sensi- the two studies which provide evidence for the effect of
tive to the recipients need state, thus suggesting a role of bonding on IA using variations of the protocol mentioned
empathy-related capabilities on reciprocal behaviours. above. It is important to have in mind although, that other
To finish this subsection, we conclude that the empir- non-human studies, despite finding evidence consistent
ical support for emotional contagion, mimicry, vicarious with IA, have failed to find any modulation of subjects’
learning and empathic concern in animals is not homoge- behaviour by the relationship quality (see Brosnan & de
nous. Simple forms of shared emotionality, such as con- Waal, 2014).
tagion and mimicry, play affiliative roles, and thus help
building a context in which altruistic exchanges could
thrive. More complex forms of empathy are yet to be IA and relationship quality
confirmed in non-humans, but they surely play impor-
In the first of the two studies to be discussed
tant roles in human altruism (Singer, 2008). All in all,
here, Brosnan et al. (2005) evaluated IA using the
sensitivity to others’ emotional states, as a factor lead-
token-exchange task with chimpanzees, Pan troglodytes.
ing to altruism, needs to be somehow constrained so that
These researchers found that refusals to exchange were
the empathetic animal avoids being exploited by cheaters.
significantly higher in the inequity than in the equity
Consistent with this view, empathy-related abilities are
condition, and that, importantly, IA was much stronger
seemingly regulated by affiliation and bonding between
in animals from colonies that had a shorter history of
individuals (Panksepp & Panksepp, 2013; Singer, 2008;
Yamamoto & Takimoto, 2012). cohabitation. The authors interpreted this last finding as
In the next subsection, we discuss IA as another affec- a possible effect of the relationship quality on IA: the
tive process that may be involved in regulating partner- longer the bond, the lesser the IA (Brosnan et al., 2005), a
ships and altruism. result similar to that observed in humans (Clark & Grote,
2003). However, these results in chimps need to be taken
with caution because years of coresidence appear to be
Fairness and IA
a very crude measure of relationship quality (see also
IA refers to the disutility or negative emotion caused by an Bräuer et al., 2009, for methodological criticism on how
outcome for the agent being unequal than an outcome for IA was assessed). As Range et al. (2012) recognised,
other individuals in a comparable situation (see Brosnan long acquaintances do not necessarily imply tolerant
& de Waal, 2014). As it is the case with empathy-related and affiliative relationships. In addition, Brosnan et al.
abilities, there is ample debate about whether non-human inferred the relationship quality effect from the compar-
animals may possess some form of IA. A major part of the ison of individuals from only three different colonies,
debate relates to experimental designs lacking appropriate which probably differed in many aspects besides years of
controls or methodological rigour to discard alternative cohabitation.
hypotheses (Bräuer, Call, & Tomasello, 2009). Despite The effect of relationship quality on IA was also
this, some authors argue that non-human IA could have assessed in dogs that lived in the same home. Range
evolved in the context of cooperation given that a negative et al. (2012) observed IA when subjects had to give their
response to being undertreated may help the individual to paw for no reward, while their partner was rewarded for
stay away from nonprofitable partnerships (Yamamoto & the same performance (in this condition, dogs underper-
Takimoto, 2012). formed compared to when both dogs were not rewarded
One of the most used protocols to study IA in for giving the paw). By asking the owners about the rela-
non-human animals consists of a task in which two tionship quality between their pets (e.g. whether they slept
individuals take turns (e.g. exchange tokens for food with in body contact), the authors found that subjects with a
the experimenter) and, while the partner is rewarded for more affiliative bond needed significantly more requests
its performance, the subject receives a higher or a lower from the experimenter to give the paw in the inequity
reward or even no reward at all after its performance condition than dogs with a more distant relationship. Sur-
instead. Under these conditions, experimenters assess prisingly, the effect of the relationship quality on dogs’
whether the subject shows signs of negative emotions, IA was in the opposite direction to that suggested by
such as rejection to participate in the task or to consume Brosnan et al.’s (2005) results in chimps, and also to
its own food, given the inequality in rewards between the results in humans (Clark & Grote, 2003). This discrep-
subject and its partner (e.g. see Brosnan, Schiff, & de ancy suggests that different methodological aspects, such
Waal, 2005; Range, Leitner, & Virányi, 2012). Despite as, for example, the operational definition of “relation

© 2015 International Union of Psychological Science

6 FREIDIN, CARBALLO, BENTOSELA

quality” or “bond,” may become crucial in evaluating this authors rarely deepen into associative learning when
phenomenon. Certainly, data are still scarce to allow sat- searching for the mechanisms involved in the origin and
isfactory conclusions, and thus, further research is needed formation of bonds. Stable bonds are typically assessed
to clarify these matters. by considering animals’ interaction patterns and their
To finish this section, and going beyond the difficulties stable preferences for interacting with an individual
in reaching strong conclusions about animals’ IA, we over other members of the group. These preferences
have shown that the potential modulation of IA by a are usually measured as the frequency or the duration
pre-existing relationship has not been a reliable finding of interactions between animals (e.g. Berghänel et al.,
in terms neither of replicability (see Brosnan et al., 2005, 2011; Fraser & Bugnyar, 2012; Sabbatini et al., 2012).
and Bräuer et al., 2009 in chimps) nor of the sign of Interestingly, such biased interactions might emerge as a
the effect (see Brosnan et al., 2005, and Range et al., product of reinforcement learning. For example, Carballo
2012, in chimps and dogs, respectively). Despite this, et al. (2015) showed that dogs developed a preference for
the preliminary results discussed above present us with a person who gave them food in the past, as opposed to a
a promising avenue of research on how long-term bonds person who withheld food from them. Dogs’ preference
may affect the expression of reciprocal behaviour which was expressed as a choice to approach and spend time
should be deepened in the future. with the donor as opposed to the other individual. In
this case, the animal may begin establishing a bond with
the preferred individual based on an associative learning
DISCUSSION mechanism. If reinforcement is maintained in the long
term, the bond could become stable, as it may happen
In this article, we discussed recent evidence in favour in dog-human interactions. Even more, if, as suggested
of the idea that some forms of animal reciprocity rely by Heyes (2012), social and asocial learning rely on
on stable social bonds. In accordance with this view, the same processes, intermittent reinforcement in social
the search for direct reciprocity has been expanded exchanges might increase persistence and stability in
from short-term contingent interactions to longer-term bonds as it does with performance in individual learning.
exchanges (Berghänel et al., 2011; Carter & Wilkinson, This is consistent with the idea of reciprocity based on
2013; Fraser & Bugnyar, 2012; Sabbatini et al., 2012). emotional book-keeping in which partial reinforcement
This affective-based reciprocity contrasts with a more may result from imperfect short-term reciprocation
calculated type which has been posed to be cognitively (Schino & Aureli, 2009).
demanding and uncommon outside humans. Indeed, this Other forms of learning could also take part in the
change in focus has led authors to explore the mecha- formation of bonds. For instance, Owren and Rendall
nisms underlying the formation and stability of social (1997, cited in Berghänel et al., 2011) hypothesised that
bonds (e.g. Yamamoto & Takimoto, 2012) and has stim- the classic conditioning could have an important effect
ulated the study of direct reciprocity in a larger group of on social relationships. According to these authors, an
species (Carter, 2014). individual’s features could become conditioned stimuli
Bonding, even outside reproductive mates and kinship, that act on its partner after repeated positive or negative
has been shown to increase evolutionary fitness (Seyfarth interactions with it. Then, the mere presence of the
& Cheney, 2012). Consistently, diverse mechanisms partner may affect the individual’s emotional state and,
have been shown to underlie bonding and altruism in therefore, trigger approach or avoidance responses. Such
animals. These mechanisms vary in their nature and learning might even result in a mental representation of
complexity, going, for example, from reflexive contagion the bond, which could in turn modulate long-term inter-
and behavioural mimicry (e.g. Palagi et al., 2014) to actions between individuals (Owren & Rendall, 1997;
prosocial behaviours based on recognition of others’ this notion is very close to Schino & Aureli’s emotional
emotions and a concern for their welfare (e.g. Singer, book-keeping). Indeed, these ideas about the role of
2008). Some authors even claim that animals may display associative learning in social behaviours could help in
IA, which may serve to avoid exploitative parternships explaining some of the results discussed in this article.
(Brosnan & de Waal, 2014; Yamamoto & Takimoto, For instance, a combination of classic conditioning and
2012). According to our review of the literature, whereas operant reinforcement could explain Bartal et al.’s (2011)
the evidence for the modulation of empathy-related results which were interpreted by the authors as if rats
abilities by prior bonding is widespread (e.g. Palagi et al., showed empathy-induced helping. According to the asso-
2014; Panksepp & Panksepp, 2013), support for the ciative learning perspective, the trapped rat, which shared
relationship between bonding and IA is still scarce and the home-cage with the free rat, could have become an
inconsistent in non-humans (see Brosnan et al., 2005; appetitive CS which thus triggered approach tendencies
Range et al., 2012). in the latter. With this incentive at sight, the free rat could
Associative learning processes can also be important have approached its trapped partner, and then have the
mechanisms in the social life of animals. Nonetheless, chance to learn the operant response to free its cage mate.

© 2015 International Union of Psychological Science

 DIRECT RECIPROCITY IN ANIMALS 7

Last, the freeing response could have been reinforced by et al. (2013) measured oxytocin in chimpanzees’ urine
re-establishing social contact between them, with no need and found that this hormone increased after grooming
to refer to empathy (Silberberg et al., 2014). Moreover, sessions with bond partners, although not when grooming
associative learning could even be behind instances of occurred outside bonds or in the absence of grooming.
short-term direct reciprocity, which otherwise tend to be This evidence does not directly speak of direct reci-
explained by presumed sophisticated cognition (Hauser procity, although it is suggestive given the recognition
et al., 2009). For example, the results of the study by that grooming is a common exchange currency in pri-
Dufour et al. (2009) on orangutans’ (Pongo pygmaeus mates (Schino & Aureli, 2009). Similar results have been
abelii) gradual acquisition of tit-for-tat reciprocity could observed when, instead of measuring endogenous levels
be interpreted along these lines. These researchers found of oxytocin, researchers exogenously administered the
a significant positive correlation between tokens given to hormone. For instance, 2 hours after oxytocin inhalation,
and received from the partner. In fact, token exchanges rhesus macaques (Macaca mulatta) increased the rewards
increased across sessions after different phases of train- given to another individual (when the alternative was to
ing, suggesting a gradual learning process. As many reward no one) relative to a control group without oxy-
other associative processes, such learning would work tocin inhalation (Chang et al., 2012). Indeed, oxytocin
more efficaciously if the events to be associated (i.e. the increased the time subjects employed observing their
succession of prosocial exchanges) happened in short partners which suggests this hormone may activate brain
time spans. In fact, evidence for reciprocity was much circuits related to social attention (Chang et al., 2012).
less convincing in a similar protocol involving longer These findings are consistent with Heyes (2012) sugges-
intervals between possible alternative exchanges (see tion that social learning does not require any special or
Amici et al., 2014). demanding cognitive ability. Perceptual, attentional and
In synthesis, associative learning might play an impor- motivational factors bias social animals towards informa-
tant role in the formation and maintenance of stable tion coming from conspecifics, which then predisposes
bonds, and, therefore, could be a mediator of reciprocal them to apply their common learning mechanisms to
behaviours. Despite this interesting possibility, to our social matters. Indeed, bonded individuals may represent
knowledge, there is no empirical evidence that directly special targets to gather information from. For example,
links learning with reciprocity. Future studies could ravens presented a stronger preference to interact with
attempt to shed light on this hypothesis, for example, an object when it had been previously manipulated by a
by experimentally manipulating the formation of bonds. close conspecific than by a more distant partner (Schwab,
A possible experimental design could involve applying Bugnyar, Schloegl, & Kotrschal, 2008).
systematic variations in the delivery of food to subjects In short, oxytocin represents a strong candidate to
when in social as opposed to individual situations. This play a role in what Schino and Aureli (2009) have called
would serve to vary the extent to which nearby con- emotional book-keeping. This affective social memory
specifics become predictive of upcoming rewards. With can be a crucial process in the stability of reciprocal
this type of protocol, experimenters could systematically
exchanges because it allows animals to track the status
vary whether animals receive more rewards in social or
of relationships without a strong cognitive burden. All in
individual situations while keeping the overall provision
all, evidence from behavioural ecology, the psychology
of rewards constant (nonetheless, experimenters would
of animal learning and the brain neurochemistry point
need to take into account dominance hierarchies and try
towards diverse socio-cognitive mechanisms that may
to avoid setups in which food can be monopolised by a
contribute to the establishment and maintenance of direct
single individual). After a period of time in which animals
reciprocity.
are predominantly rewarded either when in pairs or when
alone, subjects could take part in a task to evaluate direct
Manuscript received December 2014
reciprocity while participating with its partner or with
Revised manuscript accepted August 2015
other control subjects. If classic conditioning plays a role
in bond-based reciprocity, we would expect to see more
reciprocal exchanges in pairs that were usually rewarded REFERENCES
when together than in pairs that were not.
Amici, F., Aureli, F., Mundry, R., Sánchez Amaro, A., Mesa
Other investigations relevant to the mechanisms
Barroso, A., Ferretti, J., & Call, J. (2014). Calculated reci-
underlying social memories have been conducted on
procity? A comparative test with six primate species. Pri-
a neurochemical level. For example, oxytocin might mates, 55(3), 447–457. doi:10.1007/s10329-014-0424-4.
play an important role in allowing reciprocal exchanges Bartal, B. I., Decety, J., & Mason, P. (2011). Empathy and
through bonding. Although this neurohormone is best pro-social behavior in rats. Science, 334, 1427–1430.
known for its role in parturition and maternal care, it is doi:10.1126/science.1210789.
also involved in the processing of rewards and the estab- Berghänel, A., Ostner, J., Schröder, U., & Schülke, O. (2011).
lishment of social memories. For example, Crockford Social bonds predict future cooperation in male Barbary

© 2015 International Union of Psychological Science

8 FREIDIN, CARBALLO, BENTOSELA

macaques, Macaca sylvanus. Animal Behaviour, 81, Transactions of the Royal Society B: Biological Sciences,
1109–1116. doi:10.1016/j.anbehav.2011.02.009. 364, 3255–3266. doi:10.1098/rstb.2009.0116.
Bräuer, J., Call, J., & Tomasello, M. (2009). Are apes inequity Heyes, C. (2012). What’s social about social learning?
averse? New data on the token-exchange paradigm. Amer- Journal of Comparative Psychology, 126(2), 193–202.
ican Journal of Primatology, 71, 175–181. doi:10.1002/ doi:10.1037/a0025180.
ajp.20639. Owren, M. J., & Rendall, D. (1997). An affect-conditioning
Brosnan, S. F., & de Waal, F. B. M. (2014). Evolution of model of nonhuman primate vocal signaling. In
responses to (un)fairness. Science, 364(6207), 761–767. D. H. Owings, M. D. Beecher, & N. S. Thompson (Eds.),
doi:10.1126/science.1251776. Perspectives in ethology: Communication (pp. 299–346).
Brosnan, S. F., Schiff, H. C., & de Waal, F. B. M. (2005). New York, NY: Plenum.
Tolerance for inequity may increase with social closeness in Palagi, E., Norscia, I., & Demuru, E. (2014). Yawn contagion in
chimpanzees. Proceedings of the Royal Society B, 272(1560), humans and bonobos: Emotional affinity matters more than
253–258. species. PeerJ, 2, e519. doi:10.7717/peerj.519.
Carballo, F., Freidin, E., Putrino, N., Shimabukuro, C., Cas- Panksepp, J., & Panksepp, J. B. (2013). Toward a cross-species
sanave, E., & Bentosela, M. (2015). Dog’s discrimination of understanding of empathy. Trends in Neurosciences, 36(8),
human selfish and generous attitudes: The role of individual 489–496. doi:10.1016/j.tins.2013.04.009.
recognition, experience, and experimenters’ gender. PLoS Paukner, A., Suomi, S. J., Visalberghi, E., & Ferrari, P. F.
One, 10(2), e0116314. doi:10.1371/journal.pone.0116314. (2009). Capuchin monkeys display affiliation toward humans
Carter, G. (2014). The reciprocity controversy. Animal Behav- who imitate them. Science, 325, 880–883. doi:10.1126/
ior and Cognition, 1(3), 368–386. doi:10.12966/abc. science.1176269.
08.11.2014. Range, F., Leitner, K., & Virányi, Z. (2012). The influence
Carter, G. G., & Wilkinson, G. S. (2013). Food sharing of the relationship and motivation on inequity aversion in
in vampire bats: Reciprocal help predicts donations dogs. Social Justice Research, 25, 170–194. doi:10.1007/
more than relatedness or harassment. Proceedings of s11211-012-0155-x.
the Royal Society. Biological sciences, 208, 20122573. Sabbatini, G., De Bortoli Vizioli, A., Visalbergui, E., &
doi:10.1098/rspb.2012.2573 20122573. Schino, G. (2012). Food transfers in capuchin monkeys: An
Chang, S. W. C., Barter, J. W., Becket Ebitz, R., Watson, K. K., experiment in partner choice. Biology Letters, 8, 757–759.
& Platt, M. (2012). Inhaled oxytocin amplifies both vicarious doi:10.1098/rsbl.2012.0534.
reinforcement and self reinforcement in rhesus macaques Schino, G., & Aureli, F. (2009). Reciprocal altruism in primates:
(Macaca mulatta). Proceedings of the National Academy of Partner choice, cognition, and emotions. Advances in the
Sciences of the United States of America, 109(3), 959–964. Study of Behavior, 39, 45–69.
doi:10.1073/pnas.1114621109. Schneeberger, K., Dietz, M., & Taborsky, M. (2012). Reciprocal
Clark, M. S., & Grote, N. K. (2003). Close relationships. In cooperation between unrelated rats depends on cost to donor
T. Millon & M. J. Lerner (Eds.), Handbook of psychology: and benefit to recipient. BMC Evolutionary Biology, 12(41),
Personality and social psychology (pp. 447–461). New York, 1–7. doi:10.1186/1471-2148-12-41.
NY: Wiley. Schwab, C., Bugnyar, T., Schloegl, C., & Kotrschal, K. (2008).
Clutton-Brock, T. (2009). Cooperation between non-kin in ani- Enhanced social learning between siblings in common ravens
mal societies. Nature, 462(5), 51–57. doi:10.1038/nature (Corvus corax). Animal Behaviour, 75, 501–508.
08366. Seyfarth, R. M., & Cheney, D. L. (2012). The evolutionary
Crockford, C., Wittig, R. M., Langergraber, K., Ziegler, T. origins of friendship. Annual Review of Psychology, 63,
E., Zuberbühler, K., & Deschner, T. (2013). Urinary oxy- 153–177. doi:10.1146/annurev-psych-120710-100337.
tocin and social bonding in related and unrelated wild chim- Silberberg, A., Allouch, C., Sandfort, S., Keas, D., Karpel, H.,
panzees. Proceedings of the Royal Society, Biological Sci- & Slotnick, B. (2014). Desire for social contact, not empathy,
ences, 280, 2012–2765. doi:10.1098/rspb.2012.2765. may explain “rescue” behavior in rats. Animal Cognition, 17,
Cronin, K. A. (2012). Prosocial behaviour in animals: 609–618. doi:10.1007/s10071-013-0692-1.
The influence of social relationships, communica- Singer, T. (2008). Understanding others: Brain mechanisms
tions and rewards. Animal Behaviour, 84, 1085–1093. of theory of mind and empathy. In P. W. Glimcher,
doi:10.1016/j.anbehav.2012.08.009. C. F. Camerer, E. Fehr, & R. A. Poldrack (Eds.), Neuroe-
Dufour, V., Pelé, M., Neumann, M., Thierry, B. & Call, J. conomics: Decision making and the brain (pp. 233–250).
(2009). Calculated reciprocity after all: computation behind Amsterdam: Elsevier.
token transfers in orangutans. Biology Letters, 5, 172–175. van Baaren, R. B., Holland, R. W., Kawakami, K., & van
Edgar, J. L., Nicol, C. J., Clark, C. C. C., & Paul, E. S. Knippenberg, A. (2004). Mimicry and prosocial behavior.
(2012). Measuring empathic responses in animals. Applied Psychological Science, 15(1), 71–74.
Animal Behaviour Science, 138, 182–193. doi:10.1016/ Vasconcelos, M., Hollis, K., Nowbahari, E., & Kacelnik, A.
j.applanim.2012.02.006. (2012). Pro-sociality without empathy. Biology Letters, 8,
Fraser, O. N., & Bugnyar, T. (2012). Reciprocity of ago- 910–912.
nistic support in ravens. Animal Behaviour, 83, 171–177. Yamamoto, S., & Takimoto, A. (2012). Empathy and fair-
doi:10.1016/j.anbehav.2011.10.023. ness: Psychological mechanisms for eliciting and maintain-
Hauser, M., McAuliffe, K., & Blake, P. R. (2009). Evolv- ing prosociality and cooperation in primates. Social Justice
ing the ingredients for reciprocity and spite. Philosophical Research, 25, 233–255. doi:10.1007/s11211-012-0160-0.

© 2015 International Union of Psychological Science