Selasa, 12 Januari 2010


RECIPROCATORS AND PROSOCIALS 1 European Journal of Personality, 2001, 15, S19-S35
Individual differences and social norms: The distinction between reciprocators and prosocials
Marco Perugini
University of Essex, United Kingdom
Marcello Gallucci
Free University of Amsterdam, The Netherlands
Address correspondence to:
Dr Marco Perugini, Department of Psychology, University of Essex
Wivenhoe Park, Colchester CO4 3SQ, United Kingdom
email:; Tel. 01206 874330; Fax 0126 873590


Reciprocity is in this contribution compared with cooperation, hostility and prosociality, in order to distinguish its peculiar theoretical and empirical characteristics. Two studies are presented. Study 1 (n=166) is based on the distinction between the mechanism of reciprocity and the consequent behavior that this mechanism produces. It is shown that participants have a clear implicit theory of the personality traits underlying reciprocal behavior, and these traits are well-differentiated with respect to traits underlying cooperation and hostility. Study 2 (n=134) is based on the distinction between reciprocity as a goal and reciprocity as a strategy to achieve equality. Results show that individuals with high internalization of the norm of reciprocity allocate payoffs as a function of the valence of other's past behavior, whereas this feature is irrelevant for individuals with high prosocial orientation.
A peculiar feature of reciprocity is that in order to reciprocate, individuals might behave in completely different ways. Reciprocity can be displayed in a plethora of behaviors, as returning a favor, taking revenge of someone, exchanging proportionate goods, punishing a previously received wrong, kissing back a child, and so forth. The fact that reciprocal behavior manifests itself through actions of different nature, however, has often led scholars to equate the mechanism with the object of the mechanism. Reciprocal behavior has been interpreted as a form of cooperation, altruism, hostility, among many others actions and orientations.
A second important feature of reciprocity is that when individuals want to reciprocate, they might actually intend to achieve something else. Reciprocity is in fact a powerful mechanism to shape others' actions by delivering rewards and punishments. This makes reciprocating individuals able to achieve a remarkable number of interdependent outcomes, otherwise out of the individual's reach. Equity, cooperation, equality, and even material self-interest can be maximized by means of patterns of reciprocal behavior. This characteristic, however, has often led to the conclusion that reciprocity is a mere strategy, that is, a means to obtain a variety of goals.
Due to the aforementioned sources of confusion, and therefore to a consequently unspecific definition of the concept and mechanism of reciprocity, the large literature on reciprocity has not been very productive in the study of the personality traits and individual differences that specifically refer to reciprocal behavior in interdependence situations. Attention has been paid more to related concepts, above all cooperation and individual difference in prosociality.
This contribution seeks to investigate the unique characteristics of reciprocity in the domain of individual differences, by differentiating reciprocity from prosociality both in terms of behavior and in the corresponding personality traits. By endorsing a trait-situational perspective (Ten Berge & De Raad, in press), we investigate the concept of reciprocity
among lay people, showing that reciprocity is clearly differentiated from cooperation. Furthermore, by taking an experimental perspective, we show that individual differences in reciprocity lead to unique predictions of behavior, clearly distinguishable from individual differences in prosociality and preferences for cooperation.
The theoretical arguments upon which this differentiation is based can be outlined by addressing three basic questions: what is reciprocity, why people reciprocate and who reciprocates.
What is reciprocity?
Reciprocity is not cooperation nor it is just a consequence of having a positive attitude towards cooperation. Sometimes in the literature there has been confusion between these two concepts, especially because the concept of reciprocal altruism has been often equated with reciprocity, therefore emphasizing the positive side of reciprocity (Trivers, 1971). Reciprocity is not hostility too, even though the mechanism of reciprocity encompasses also various forms of hostile and negative behaviors. Reciprocity is a mechanism that can lead to positive and negative forms of behavior depending on the specific context eliciting a reciprocal action. This conditional nature of reciprocity was expressed by Gouldner first (1960) and elaborated upon with the concept of Tit-for-Tat (Axelrod, 1984; & Dion, 1988). The emphasis has been generally placed on the strategic value of reciprocity as a norm able to increase future payoffs and deter defection, given some structural features of the interaction. One of these structural features that has been traditionally emphasized is the presence of repeated interactions between the individuals (Axelrod, 1984; Kreps, Milgrom, Roberts, & Wilson, 1982). However, as we shall argue in the following pages, recent experimental results have clarified that the presence of repeated interactions is not a necessary condition to observe reciprocal behavior among some participants (Gallucci & Perugini, 2000; Van Lange, 1999). Reciprocity, therefore, is not always a strategy to maximize self-interest.
We maintain that reciprocity is a conditional behavior aimed at reacting to a behavior with another behavior of the same valence (Perugini & Gallucci, 1998; Gallucci & Perugini, 2000 ). Reciprocity is clearly differentiated from cooperation and hostility because it is conditional and it matches actions according to their valence: A positive actions is reciprocated with positive actions, a negative action with negative ones.
What remains unclear from this analysis is whether these differences have actual correspondence in real life. Do people differentiate between reciprocators, cooperators, and hostile persons? And if so, what kind of criteria are they using? Is there any implicit conception of these three personality dimensions used by lay people?
Why people reciprocate?

Obviously, a specific instance of reciprocal behavior can be performed for a variety of reasons and by a variety of persons and it can be explained from different perspectives. Reciprocity has been observed in living beings with undoubtedly different motivations and personalities, such as vampire bats (DeNault, & McFarlane, 1995), hyenas, (Trivers, 1971), apes (de Wall, 1992), humans of different ages and cultures (Goldstein, Field, & Healy, 1989; Patterson & Bettini, 1993; Windle, 1994), a result that by itself prompts for different possible answers to our present question.
Striving for self-interest has been very often suggested as a motivational basis of reciprocal behavior (Axelrod, 1984; & Dion, 1988). When the interdependent situation is repeated over time, reciprocal behavior is not exploitable by more selfish strategies and it can profit from mutual cooperation. This strategic value has been supported by formal analyses (Kreps, Milgrom, Roberts, & Wilson, 1982), computer simulations (Axelrod, 1984), and experimental results (Hoffman, McCabe & Smith, 1998; Komorita & Parks, 1999).
A second explanation has been provided within the tradition of interdependence theory (Kelley & Thibaut, 1978). In this framework, individuals are expected to have different preferences concerning other's outcomes in situation of interdependence of outcomes.
, tending to maximize the relative advantage of their outcomes over others’ outcomes. Research has shown that prosocials approach situations of interdependence in a cooperative manner and continue to do so until the other fails to cooperate (McClintock & Liebrand, 1988; Sattler & Kerr, 1991). Recently, Van Lange (1999) has proposed an integrative model of social value orientations where the traditional emphasis on the maximization of joint outcomes in defining prosocials has been integrated with the assumption that they are also motivated to achieve equality in outcomes. This latter conceptualization can be used to explain and predict some forms of reciprocal behavior also in not-repeated interactions. In other words, prosocials are assumed to strive also to restore equality of outcomes and this can lead in turn to reciprocate their partner's behavior such that this equality is achieved (see also De Cremer and Van Lange, 2001). Interdependence theory assumes that the given matrix representing the mutual outcomes of the interaction depending on the combinations of behaviors is actually transformed by people in the effective matrix, whereby individuals transform the given outcomes according to their social orientations. Social orientations have the quality of a personality dimensions, in that it is assumed that people have stable and reliable preferences that guide their behavior in situations of interdependence. Among the several possible transformations, attention has been devoted to three of these “personality types”: prosocial, who are tending to maximize outcomes for both themselves and others and minimize the differences, individualists, tending to maximize their own outcomes regardless of others’ outcomes, and competitors
We have adopted a wider perspective (Perugini & Gallucci, 1998). By taking a pluralistic view on motivation, we have argued that reciprocity can be defined both as a strategy and as a goal itself. This view is based on the analysis of reciprocal behavior in terms of means and ends. Reciprocity can be instrumental to achieve a specific goal, and actors aware of its effects can employ it accordingly. Thus, a rational actor that is able to foresee the long term effect of reciprocity on her outcomes, can act in a reciprocal way because she
believes that this is the best way to gain as much as possible (Axelrod, 1984). Likewise, a prosocial actor who wants to achieve equitable allocations of resources can employ a reciprocal behavior in order to balance the amount of resources that are exchanged in the situation (Van Lange, 1999).
However, it is possible to identify a type of reciprocity whereby behavior and its underlying motivation are convergent and reciprocity can be defined as a goal: reciprocity as an internalized norm. A key feature of this definition is that internalized reciprocity should be viewed as an individual preference, since it represents a personal tendency to behave in a reciprocal way, regardless of strategic considerations. Therefore, the underlying motivation sustaining a reciprocating behavior can be endogenous rather than functional to achieve some alternative goal. Reciprocity can be the goal per se. This distinction echoes a general distinction between social and personal norms (Cialdini, Kallgren, & Reno, 1991; Etzioni, 1989; Kerr, Garst, Lewandowski, & Harris, 1997; Schwartz, 1977), whereby the latter can be conceived as social norms that become internalized via social learning and acquire a motivational drive on their own. Indeed, individuals holding internalized social norms (or personal norms) are likely to conform to their dictates even when they are not observed or sanctioned externally, through the anticipation of self-punishments and rewards based on their own internalized values (Schwartz, 1977).
The distinction between reciprocity as a mean and reciprocity as a goal has two crucial consequences: First, it is possible to clarify what situational and interpersonal features are necessary for certain behaviors to be acted. Second, it is possible to clarify what traits are necessary for a particular form of reciprocal behavior to be acted.
The first consequence is well exemplified by considering reciprocal behavior when instrumental for self-interest. This behavior should be acted only when future profits are available, they can be obtained via reciprocal behavior and actors are able to foresee this possibility. As we mentioned earlier, experimental results have shown that reciprocity is
observable when none of these conditions are satisfied. Using situations where no repetition was possible, and the reciprocal strategies were always outperformed by unconditional selfishness, it has been shown that a fair proportion of the experimental sample employed reciprocity as the behavioral rule to allocate resources in various interdependent situations (Bolton, & Zwick, 1995; Gallucci & Perugini, 2000; Gallucci & Perugini, 2001). These results clearly suggest that whereas material self-interest can explain some instances of reciprocal behaviors, it cannot be the sole underlying motivation of reciprocity.
It is equally possible to specify the situational features that are necessary for prosociality to explain reciprocal behaviors. Prosociality leads individuals to reciprocate because prosocials prefer equitable outcomes over advantageous outcomes (Van Lange, 1999). Thus, prosocials should be concerned only with the outcomes of their behavior and should react only to the other's allocation outcome. On the other hand, the norm of reciprocity concerns the reaction to the valence of others' behavior, with the outcome per se being only one component of the overall valence. Thus, prosociality should lead to reciprocal equality, whereas the internalized norm of reciprocity should lead to reciprocal behavior even when equality in the outcomes is not the consequence of the action.
This latter analysis enables us to outline a further distinction, which brings us to the core of this contribution: the distinction between the effects of individual differences in the internalized norm of reciprocity and prosociality.
Who reciprocates?

The definition of reciprocity as an internalized norm has been encapsulated in a formal model developed using game theory (Perugini & Gallucci, 1998; Gallucci & Perugini, 2000). A basic assumption of this model of reciprocity is that there are individual differences in abiding by the prescription of the norm. In their experiment, as well as in other experiments (Gallucci & Perugini, 2001; Perugini & Gallucci, 2001), it has been found that some individuals systematically do reciprocate whereas others do not. The question therefore
becomes who are those individuals who reciprocate even when there are no material or social advantages in so doing.
According to Van Lange (1999), prosocials, as defined by means of the Social Value Orientation triple-dominance measure (Kuhlman & Marshello, 1975), show evidence of reciprocal behavior (also called behavioral assimilation) even in single-trial prisoner’s dilemma. Prosocials’ level of cooperation was found to be systematically influenced by the partner's cooperation, with more cooperation following a partner's choice of cooperation. Therefore, in this line of thinking, reciprocal behaviors can be considered as a result of being a prosocial, which in turn implies, as re-defined by Van Lange (1999), being motivated to both maximizing joint outcomes and achieving equality in outcomes. This latter goal can be achieved both through punishments and rewards and, depending on the specific situation, can therefore lead to predict reciprocal behavior.
According to our analysis, reciprocity can be a goal on itself, and thus reciprocal behavior is not limited to the purpose of equality. If one considers reciprocity as a personal norm, then it is conceivable that individuals will differ directly in how much they are willing to follow the norm. In other words, it can be argued that there are reliable individual differences in the motivation to reciprocate per se rather than on some other personality dimensions (e.g., prosociality) that in turn would lead to reciprocate. Indeed, Perugini and Gallucci have recently developed such an individual difference measure, called Personal Norm of Reciprocity (PNR) scale, and validated it successfully in a series of studies (Gallucci & Perugini, 2001; Perugini & Gallucci, 2001; Perugini, Gallucci, Presaghi, & Ercolani, 2001). The PNR measures three dimensions: positive reciprocity, focusing on positive reactions to positively valued behaviors, negative reciprocity, focusing on negative reactions to negatively valued behaviors, and beliefs in reciprocity, concerning the personal belief that both forms of reciprocity are widely adopted by many people and generally effective in interpersonal and social interactions (Perugini et al., 2001). We argue that these
constructs of reciprocity should have a main role in predicting individual differences in reciprocal behavior, especially when it is a consequence of a personal motivation rather than of material convenience.
Reciprocators are not (necessarily) prosocials

Even though there is some empirical evidence showing that prosocials may reciprocate even when there is no obvious strategic convenience (Van Lange, 1999; De Cremer & Van Lange, 2001), we argue that in the general case reciprocators are not prosocials. The underlying mechanism of reciprocity is to react to someone else action with an action of the same valence, such as positive behaviors are rewarded and negative behaviors punished. The main motivation of reciprocators therefore is to sanction someone else behavior, including also behaviors which are expected rather than already acted. Their concern therefore is not so much to restore equality in outcomes, but to match the valence of others' behavior with equivalent behavior. Therefore, the general case is that the mechanisms of reciprocity and of prosociality are different and lead to different behaviors, even though there can be situations whereby, due to the nature of the payoffs and the structure of the interaction, the empirical predictions can be equivalent. Of course, the first key difference is linked to the performance of negative behaviors. Whereas reciprocators, as defined by high scores in the scale of negative reciprocity, can easily resort to forms of negative behaviors (e.g., punishment) if the situation requires (e.g., the other behaved negatively), prosocials are expected to be less prone to act in negative terms (although they might do so, cf. Van Lange, 1999). In other words, prosociality mainly pertains to positive forms of behaviors (e.g., cooperation), whereas reciprocity includes both positive and negative behaviors. A second feature differentiating prosocials from reciprocators is the importance of the valence of someone else behaviors. As a consequence of their reactivity to others’ actions, reciprocators are expected to be very sensitive to the valence of the action and to take into account factors such as its intentionality. On the other hand, there is nothing in the definition of the prosocials
concerning these issues.
The present contribution

This contribution presents two studies. The first study endorses a trait-situational perspective (Ten Berge & De Raad, 1999, in press) to investigate the concept of reciprocity among lay people. The focus is on individuating those trait-situation features that can distinguish between reciprocity, cooperation and hostility. The second study develops further some of the findings of the first study, and focuses on the differential effects of the norm of reciprocity and prosociality on behavior towards others.
Study 1

Different traits can lead to the same behavior. Without a careful study of the situational contingencies within which behaviors are performed the risk is to fail to appreciate this if-then connection (Wright & Mischel, 1987). Moreover, not all personality characteristics are expressed in all situations. As argued by Ten Berge and De Raad (1999), certain traits can be expressed preferentially in certain situations. Reciprocity is one of such traits. Suppose that Mary observes John acting very kindly towards Jim: she would be tempted to conclude that John is cooperative. However, if soon afterwards Carolyn sees John acting unkindly towards David she might arrive at the opposite conclusion: John is a hostile person. One way to solve this apparent inconsistency is to look for the situational contingencies characterizing John’s behavior. If Mary and Carolyn would have seen that, say, Jim was behaving very nicely with John whereas David was behaving very rudely, then they might have come to the conclusion that John is a reciprocator. In fact, reciprocity is leading to both kind and unkind behaviors depending on the previous history of the interaction. The valence of the behavior performed by the other in the interaction is the key factor that is assumed to shape the reciprocal reaction. On the other hand, cooperation and hostility represent forms of relatively unconditional behavior. Cooperators and hostile persons are assumed to behave nicely and rudely, respectively, in a generalized way and without a strong modulation of their behavior
as a function of the previous history of the interaction and should therefore be less sensitive to the valence of someone else previous behavior. Obviously, it is likely that there will be limits to this inclination to behave kindly or unkindly, but these limits will be typically biased in the direction of the personality disposition (e.g., cooperators will tend to cooperate even in conditions where average people will not do so).
In this study we investigate whether lay people actually differentiate between reciprocity, cooperation and hostility and, if so, whether they use the same principle of behavior conditionality that we have sketched so far.
Participants and procedure.

One-hundred-sixty-six participants were recruited at the University of Rome. Approximately one third of the sample was composed of friends, parents or relatives of undergraduate students, who composed the remaining two thirds of the total sample. Fifty-seven were males and 109 females, with an average age of 28.3 years (SD=12.2). They were given a booklet, containing also some unrelated questionnaires, where they were asked to write at least 1 situation for each of the trait reciprocity, cooperativeness, and hostility. The instructions were a slight modification of those used by Ten Berge and De Raad (in press). In particular, after a brief presentation of the aim of the study, participants were encouraged to indicate situations and behaviors that would demonstrate that somebody has a given personality trait. They were asked to use the form “When [event] [behavior]”. Some examples concerning an unrelated personality trait (recklessness) were provided (e.g, “When s/he drinks, s/he does not bother if after s/he has to drive the car”) in order to familiarize the participant with the task. Participants were encouraged to write at least one and in any case no more than 10 of these sentences for each of 3 traits, limiting their time to no more than 5 minutes for each trait. The instructions were followed by three pages containing space for up to 10 sentences for each trait. The personality dimensions were simply introduced as
“reciprocator- somebody who reciprocates”, “cooperator – somebody who cooperates”, and “hostile – somebody who is hostile”, in order to rely on participants’ implicit theory rather than providing a more exhaustive but possibly biasing specific definition of each trait.
Coding of the sentences and hypotheses. The sentences were coded by two independent raters according to three criteria. First, the part concerning the stem [event] was coded in terms of valence, negative vs neutral vs positive. The same coding in terms of valence was done on the stem [behavior]. Finally, the unit [event] [behavior] was coded in terms of action congruence, different vs same, that is whether the behavior was congruent with the event. The frequencies of these three criteria for the three dimensions of reciprocity, cooperativeness and hostility represented the dependent variables.

The first set of hypotheses concerns the valence of the action. We expect that the valence of the event and the behavior would tend to be similar for reciprocity. The mechanism of reciprocity prescribes to return good for good and bad for bad (Gouldner, 1960). Therefore, we expect that this conditional link would be reflected in a very high frequency of sentences whereby a positive event is followed by a positive behavior and a negative event by a negative behavior. On the contrary, we do not expect such symmetry for cooperativeness and hostility, given that these two traits are reflecting relatively unconditional dispositional inclinations. Moreover, we would expect prevalence of positive [behavior] for cooperativeness and of negative [behavior] for hostility, regardless of the stem [event]. In particular, we also expect that many neutral [event] will be followed by positive [behavior] for cooperativeness and by negative [behavior] by hostility. In other words, we expect that the personality dispositions of cooperativeness and hostility will be reflected in a relatively unconditional preference for positive or negative behavior.
The second hypothesis states that there will be action congruency for reciprocity but not for cooperativeness and hostility. The reason is that reciprocity involves giving back what has been received and therefore the congruence between type of event and type of behavior is
more important. Even though it is possible to reciprocate in different kinds, typically the easiest way is to return the same or a very similar action to what it has been previously received, given that this would guarantee equivalence of valence between the other’s behavior and the agent’s reaction (Batson, 1993). On the contrary, for both cooperativeness and hostility this is not a concern, as both dispositions are relatively unconditional and therefore should not be much affected by this action congruency principle.
Results and discussion
Participants produced 679 sentences, with an average of 4.09 each. Of these, 79 were not valid, most typically because participants failed to write the sentences as an if-then unit, leaving a total of 600 valid sentences. The agreement between the two coders was high1, with Cohen’s kappa of .80, .93, and .74 for [event], [behavior], [event]-[behavior] congruence, respectively. Disagreements were solved through discussion. The sentences were so distributed: 175 for reciprocity, 185 for cooperativeness, and 240 for hostility. The first hypothesis can be simply checked by ascertaining the correlation between the [event] [behavior] valences separately for each disposition. The Spearman’s rho correlation coefficient in the case of reciprocity was highly significant and equal to .81, therefore indicating a strong positive relation between the two valences. For both cooperativeness and hostility the value was low and not significant (.02 and -.00, respectively), meaning that for these two dispositions there was no conditional link between [event] and [behavior]. The cross-tabulations of the [event] [behavior] valences can give more fine-grained information and are relevant to the second part of the first hypothesis (Table1). These can be transformed in conditional probabilities, which are expressing the probability of a specific [behavior] having a certain valence given the valence of an [event] (Table 2).
[Insert Tables 1 and 2 about here]
The hypotheses are confirmed by the conditional probabilities. For cooperativeness there is a strong bias for positive behaviors, regardless of the valence of the event. In fact, the
probabilities of a positive behavior occurring after a negative, neutral, or positive event are all higher than .90 and very similar to each other, meaning that whichever the event, the behavior is very likely to be positive. An opposite pattern can be found for hostility, with high probability of occurrence of a negative behavior regardless of the valence of the event. Indeed, also in this case the relevant conditional probabilities are very high and approaching unity. The results are more articulated for reciprocity. When an event is negative, the conditional probability of a negative behavior is very high (.954), whereas when it is positive, the probability of a positive behavior occurring afterwards is also very high (.972). However, when the event is neutral, apparently there is a slight bias towards a positive behavior (.667) rather than a neutral one (.333). It has to be noted however that the raw cross-frequencies are very low (6 and 3, respectively) and therefore it remains to be seen whether this result would generalize to a bigger sample of sentences. In general, the results lend full support to our first set of hypotheses.
The second hypothesis involves comparing the distribution of action congruency between reciprocity and the null hypothesis of equal distribution, and between the three dispositions (Table 3).
[Insert Table 3 about here]
The units [event] [behavior] generated for reciprocity tend to be more composed of the same actions than of different actions (75.4% vs 24.6). The proportion is significantly different from an equal distribution (Z=6.72, p<.001). On the contrary, for both cooperativeness and hostility the actions tend to be more different than the same (75.1% and 74.2% of different actions for cooperativeness and reciprocity, respectively). The proportion of same actions is significantly different when comparing reciprocity with cooperativeness (Z=9.58, p<.001) and hostility (Z=10.00, p<.001)., whereas the latter two dispositions do not differ from each other (Z=0.21, p>.05).
The results of the first study have shown that reciprocity as a personality disposition
has some specific features. Unlike cooperativeness and hostility, reciprocity has a conditional connotation that emphasizes the matching between what has happened to an agent and what the agent is going to do as a consequence. This symmetry is equally reflected in positive action following positive behaviors and negative actions following negative behaviors. Both cooperativeness and hostility instead present a behavioral bias in the direction congruent with the personality disposition. For cooperators, positive actions are performed regardless whether these were preceded by negative, positive or neutral behaviors, whereas for hostile persons the reverse is true. Moreover, only reciprocators appear to be concerned with reacting in the same way as the action and follow therefore an action congruency principle. This investigation in the lay people conceptions of these three personality dispositions supports therefore the theoretical frame that we have sketched in this contribution. One could argue that results may have been different if we had used the concept of prosociality instead than cooperation2. It is possible that the former concept would have been less suggestive of a bias towards positive actions. However, it is also the case that whereas the concept of cooperation is often used in everyday language, the concept of prosociality is more technical and less likely to be used by lay people. Nonetheless, we believe that the adoption of a broader spectrum of concepts, such as prosociality, linked to positive and negative interpersonal actions would be an interesting focus for future research.
Study 2
The second study investigates further the role of valence for reciprocators. One of the key assumptions of the model of reciprocity is that reciprocators react to the valence of the action and not merely to its material outcome (Gallucci & Perugini, 2000; Perugini & Gallucci, 1998). In other words, reciprocators are expected to be particularly sensitive to factors such as the intention of the interacting agent whereas they should not be particularly influenced by strategic considerations such as the expectation of a repeated interaction or the status of the interacting agent (potential partner vs friend). Some evidence supporting these
claims is already available. Reciprocators, as defined by high scores in the Personal Norm of Reciprocity scale, showed to be particularly sensitive to whether somebody did behave negatively or positively whereas they where not influenced by whether the interaction was expected to be repeated or one shot (Perugini & Gallucci, 2001). There is no rationale instead to hypothesize a similar mechanism for prosocials. It is not clear in fact why factors such as intentionality or valence should affect a prosocial in his/her allocation decisions.
The aim of the second study is therefore to test whether reciprocators are particularly sensitive to the valence of someone else behavior, whereas prosocials and individualists are not, and relatively insensitive to the type of relationship. The study involves hypothetical choices without material payoffs. We expect consequently that due to the lack of monetary incentives, participants as a whole will show more generosity than they would should real money be at stake. Therefore, the structure of the situation is such that negative reciprocity will be especially important, given that the increased average payoff allocation would likely not leave any space to detect influences due to positive reciprocity. However, previous results have shown that both dimensions of reciprocity are able to predict deviations from economic rationality (i.e., maximization of personal payoffs) when real payoffs are involved. Indeed, Gallucci & Perugini (2001) found that participants with high scores in both positive and negative reciprocity were willing to ask for strategically irrelevant information, instead than strategically relevant, and reciprocate at a personal cost the other’s behavior in a sequential double-dictator game. Likewise, Gallucci & Perugini (2000) found in the so-called Reciprocity Game that positive reciprocators, as defined by high scores in an ad hoc scale that can be considered as a proxy of the positive reciprocity dimension, were willing to sacrifice part of their endowment in order to reward the other agent after a previous positive interaction.
Participants and procedure.

The total sample was composed of 134 students of the University of Leicester, 23 males and 111 females, with an average age of 21.3 (SD=6.2). Participants were given a self-paced booklet containing a series of measures. Two personality measures were included in the material.
a) Social Value Orientation (SVO). We adopted the triple-dominance measure of social values (Kuhlman & Marshello, 1975) with nine decomposed games (see Van Lange, De Bruin, Otten, & Joireman, 1997 for details and the scoring procedure). On the basis of their choices, participants can be classified as prosocials, individualists, and competitors. If the pattern of choices is inconsistent, respondents are not classified. Participants were classified as 40 prosocials, 67 individualists, 10 competitors and 17 not classifiable. As the number of competitors was too low, we decided to consider in the remaining of the analyses only prosocials and individualists3. The sample for the analyses was therefore composed of 107 participants, 19 males and 88 females, with an average age of 20.8 (SD=5.7).
b) Personal Norm of Reciprocity (PNR) questionnaire. This questionnaire is composed of 27 items measuring three dimensions (9 items each), positive reciprocity, encompassing positive reactions to positively valued behaviors, negative reciprocity, where the emphasis is on matching negative reactions to negatively valued behaviors, and beliefs in reciprocity, concerning the belief that both forms of reciprocity are widely adopted and generally effective. The questionnaire has been validated in two countries, Italy and United Kingdom, in a series of studies reported in Perugini et al. (2001).
Besides the two questionnaires, the booklet contained four sets of six allocation matrices in random orders, but fixed for all participants. Each matrix included three possible allocations of payoffs: equitable, advantageous, and disadvantageous for the participants. Each set of matrices was introduced manipulating the other’s previous behavior (valence,
positive vs. negative) and the type of relationship (friendship vs expectation of future relationship), according to a 2 x 2 within-subjects factorial design. Participants were asked to choose which allocation they preferred among the three options for each of the six matrices. The dependent variable was the amount of average payoff across the six matrices allocated to the other, therefore ranging from 98.33 to 498.33. For instance, the first allocation matrix for the combination positive valence and expectation of future relationship was as follows:
The other person is someone whom you do not really know, but you are going to meet very often in the future [expectation of future relationship], and who recently behaved nicely with you [positive valence]. For each of the following six choice situations, circle A, B, or C, depending on which column you prefer most.
(1) YOU 480 540 480
OTHER 80 280 480
For simplicity reasons, the dependent variable has been divided by the maximum possible payoff allocated to the other and multiplied by 100, therefore transforming the value in percentage of allocated payoff relative to the maximum allocation to the other.
The first hypothesis is that the allocation choices will be affected by the valence of the behavior. In particular, we expect that higher allocations will be given to hypothetical persons who behaved nicely previously than to the ones who behaved rudely. This prediction is straightforwardly derived from the norm of reciprocity and it has been already supported in another study (Perugini & Gallucci, 2001). The second hypothesis is that the type of relationship will also influence the allocation, with more payoff allocated to a friend than to somebody with whom one is expected to interact many times in the future, therefore only with the potential of becoming a friend. However, this influence should be particularly relevant when the other behaved rudely, due to consideration of possible future positive returns if the relationship is going to go in a virtuous cycle. Hence, participants might feel inclined to partly forgive the initial negative behavior for the sake of establishing a positive pattern in the future. The third hypothesis therefore states that there will be a significant
interaction between valence and type of relationship.
Most interesting for this contribution is the role of personality dimensions, given that we expect an influence of reciprocity on the allocation. However, given the nature of the experimental situation, we expect that negative reciprocity is especially influential and this should result in a significant interaction. In fact, Study 1 has shown that reciprocity, unlike prosociality, is characterized by a conditional if-then action whereby the valence of someone else action is matched. Therefore, the fourth hypothesis is that individual differences in reciprocity will interact with the valence of the behavior and this will be primarily due to negative reciprocity, which should significantly affect allocation choices when the other behaved negatively, regardless of whether s/he was a friend or a potential one. Three ancillary hypotheses are also worth mentioning, even though strictly speaking two of them cannot be defined as such given that they rely on accepting the null hypothesis. Prosociality is expected to have a main effect on allocation choices, with prosocials giving more payoff than individualists. However, it is not expected to have a significant interaction with valence, as the dimension prosocial-individualistic is not expected to be differentially influenced by what the other did in the past. Likewise, but for different reasons, reciprocity is not expected to interact with the type of relationship. In fact, if reciprocity as measured by the PNR scales is primarily an internalized norm, it should not be particularly sensitive to factors such as the type of relationships, which have a more strategic nature. In other words, the reaction of a reciprocator can and should be affected by the valence of someone else action, as this is a crucial element (cf. Gallucci & Perugini, 2000), but it should be relatively unaffected by elements which do not bear strong theoretical relevance, such as the type of relationship. If, contrary to our theoretical analysis, reciprocity is primarily a strategy supported by maximization of material outcomes, then the type of relationship would be affecting the agent’s choice, as it could be construed as an indicator of how much is likely that positive future returns will follow the agent’s choice (e.g., friends are more likely to give future
positive returns than potential partners).
Results and discussion

To test the set of hypotheses a mixed ANCOVA 2 (Valence, positive vs negative) x 2 (Future interdependency, present vs absent) x 2 (SVO, prosocial vs. individualist) and using the scores in the three reciprocity scales as covariates was performed. This approach was preferred over the more traditional possibility of dichotomizing the scores in the scales, by creating groups of high and low scorers, as this latter results in reduced statistical power and biased estimates (Maxwell & Delaney, 1993; McClelland & Judd, 1993), which would be even more likely in the present case given the moderate sample size.
There was a main effect for Valence (F(1,102)=174.910, p<.001) and for Type of relationship (F(1,102)=66.424, p<.001). More payoff was allocated when the other behaved positively (94.81) than negatively (68.89), and when the interaction was with a friend (86.75) instead than with a potential partner (76.94). The interaction between the two factors was significant (F(1,102)=30.968, p<.001). As depicted in Figure 1, it was mostly due to a much stronger impact of the type of relationship when the behavior was of a negative valence (76.78 with a friend vs 60.99 with a potential partner, difference of 15.79) than when it was of a positive valence (96.73 with a friend vs 92.89 with a partner, difference of 3.84). [Insert Figure 1 about here] These results are therefore supporting the first three hypotheses. Concerning the role of the scales of reciprocity, negative reciprocity did have a significant covariate effect (F(1,102)=8.196, p<.01). However, this was qualified by a significant interaction with the Valence (F(1,102)=9.385, p<.001) 4. An inspection of this interaction showed that it was due to a significant decrease in the allocation of the payoffs with increasing scores in negative reciprocity when the valence of the behavior was negative (β=-.37, p<.000) but not when it was positive (β=-.00, p=.962). The fourth hypothesis was therefore fully confirmed. All other ancillary hypotheses were also supported. The SVO type did have a significant main effect RECIPROCATORS AND PROSOCIALS 22 (F(1,102)=18.505, p<.001), meaning that prosocials were allocating more payoff (88.71) than individualists (77.75). However, this main effect was not qualified by any interaction with the other factors, including the valence of the behavior. Finally, none of the scales of reciprocity interacted with the factor type of relationship5. General discussion The two studies presented in this contribution investigated some of the features of reciprocity as a personality dimension. The first study showed that reciprocators are expected to act conditionally on someone else behavior, unlike cooperators and hostile persons, and tend to return kind for kind via the action congruency principle. The second study showed that reciprocators take particularly into account the valence of someone else behavior, unlike prosocials and individualists, and react by giving less payoffs to the ones who behaved rudely with them, regardless of the type of relationship characterizing their interaction. These two studies, together with other recent studies (Gallucci & Perugini, 2000, 2001; Perugini & Gallucci, 2001; Perugini et al., 2001), are starting to give a more precise characterization of the mechanism of reciprocity and to specify what kind of individual differences modulate this mechanism and how they work. First, the focus is on reciprocity as an internalized mechanism, emphasizing therefore the link between motivation to reciprocate and reciprocal behaviors. Second, a distinction between different but related personality characteristics linked to reciprocity (positive reciprocity, negative reciprocity, beliefs in reciprocity) is introduced. Third, reciprocators are distinguished from hostile persons and from cooperators or prosocials. This last distinction deserves some more comments. Besides the theoretical rationale that we have developed in this contribution and the results of the second study, there is also correlational evidence clearly supporting this distinction (cf. Perugini et al, 2001, Table 3). The SVO and the PNR have been administered to a sample of 146 Italian and 226 English participants. Participants were classified as prosocial or proself (including individualists and competitors, cf. Liebrand & Van Lange, 1991). In Italy, this RECIPROCATORS AND PROSOCIALS 23 reduced SVO was not related to the overall score in the PNR scales, whereas it was related with the PNR scale of positive reciprocity (r=.24, p=.004), meaning that higher scores in the scale were associated with being classified as prosocial. Interestingly, a different pattern emerged in the United Kingdom, with a slight negative correlation with the overall score (r=-.17, p=.011) mainly due to a stronger negative correlation with the PNR scale of negative reciprocity (r=-.28, p<.001). English cooperators tended to have lower scores in the negative reciprocity scale, but not higher scores in the positive reciprocity one (r=.06, p=.34). It appears therefore that there is a modest association between being a prosocial and being a reciprocator and clear evidence of discriminant validity. Furthermore, whether this relation is on the positive or on the negative side of reciprocity depends on the socio-cultural context. These results suggest that prosocials are more likely to reciprocate positive behavior (e.g., to cooperate after the other person cooperated) in Italy, whereas they are less likely to reciprocate negative behavior (e.g., to defect after the other defected) in England, and imply that to cooperate more is not the same as to defect less. Of course, when the choices are dichotomous, to cooperate is equivalent to not defect, but in real life often choices come in degrees and individuals have the opportunity to modulate their behaviors by deciding to cooperate or to defect to a given extent. It is possible that in more realistic conditions the asymmetry between increased cooperation and decreased defection can emerge and can be differentially associated to the different sides of reciprocity, depending on the prevailing norm in a given society concerning the social desirability of reacting negatively or positively to previous behaviors. These speculations should be tested in future research. This contribution has some obvious limitations, especially concerning the second study where no real money was involved. However, the pattern of results is consistent with other studies with monetary incentives (Gallucci & Perugini, 2000, 2001). Moreover, the results of the two studies delineate a consistent pattern characterizing reciprocators and distinguishing them from cooperators or prosocials. We started this contribution claiming that reciprocity is RECIPROCATORS AND PROSOCIALS 24 not cooperation and reciprocators are not prosocials. We can end it with the same claim. RECIPROCATORS AND PROSOCIALS 25 References Axelrod R, Dion D. 1988. The further evolution of cooperation. Science. 242: 1385-1390. Axelrod R. 1984. The evolution of cooperation. Basic Books: New York. Batson CD. 1993. Communal and exchange relationships: What is the difference? Personality and Social Psychology Bulletin. 19: 677-683 Bolton GE, Zwick R 1995. Anonymity versus punishment in ultimatum bargaining. Games and Economic Behavior. 10: 95-121. Cialdini, RB, Kallgren CA, Reno, RR 1991. A focus theory of normative conduct: A theoretical refinement and reevaluation of the role of norms in human behavior. In MP Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 24, pp. 201-234). New York: Academic Press. Cialdini RB, Trost MR. 1998. Social influence, social norms, conformity and compliance. In DT Gilbert, ST Fiske, G Lindzey (Eds). The handbook of social psychology, Vol. 2 (4th ed.). (pp. 151-192). McGraw Hill: Boston, MA. De Cremer D., Van Lange PAM. 2001. Why Prosocials Exhibit Greater Cooperation than Proselfs: The Roles of Social Responsibility and Reciprocity. European Journal of Personality, this issue. DeNault LK, McFarlane DA. 1995. Reciprocal altruism between male vampire bats (Desmodus rotundus). Animal Behaviour. 49: 855-856. De Waal, F 1992. Coalitions as part of reciprocal relations in the Arnhem chimpanzee colony. In Harcourt & F de Waal (Eds), Coalitions and alliances in humans and other animals (pp. 223 258). Oxford University Press: Oxford. Etzioni A. 1989. The moral dimension: Towards a new economics. New York: Free Press. Gallucci M, Perugini M. 2000. An experimental test of a game-theoretical model of RECIPROCATORS AND PROSOCIALS 26 reciprocity. Journal of Behavioral Decision Making, 13: 367-389. Gallucci M., Perugini M. 2001. Reciprocity and information seeking. Unpublished manuscript. Gouldner AW 1960. The norm of reciprocity: A preliminary statement. American Sociological Review, 25: 161-178. Hoffman E, McCabe K, Smith V. 1998. Behavioral foundations of reciprocity: Experimental economics and evolutionary psychology. Economic Inquiry, 36: 335-352. Kelley, HH, Thibaut, JW. 1978. Interpersonal relations: A theory of interdependence. New York: Wiley. Kerr, NL, Garst, J, Lewandowski, DA, Harris, SE. 1997. That still, small voice: Commitment to cooperate as an internalized versus a social norm. Personality and Social Psychology Bulletin, 23: 1300-1311. Komorita SS, Parks CD. 1999. Reciprocity and cooperation in social dilemmas: Review and future directions. In Budescu, DV. Erev I, Zwick R. (Eds.). Games and human behavior: Essays in honor of Amnon Rapoport (pp. 315-330). Lawrence Erlbaum Associates: Mahwah, NJ. Kreps DM, Milgrom PR, Roberts J, Wilson, RJ 1982. Rational cooperation in the finitely repeated prisoners' dilemma. Journal of Economic Theory, 27: 245-252. Kuhlman DM, Marshello A. 1975. Individual differences in game motivation as moderators of preprogrammed strategy effects in prisoner’s dilemma. Journal of Personality and Social Psychology. 32: 992-931. Lindbeck A. 1995. Welfare state disincentives with endogenous habits and norms. Scandinavian Journal of Economics. 97: 477-494. Maxwell SE, Delaney HD. 1993. Bivariate median splits and spurious statistical significance. Psychological Bulletin. 113: 181-190. McClelland GH, Judd CM. 1993. Statistical difficulties of detecting interactions and RECIPROCATORS AND PROSOCIALS 27 moderator effects. Psychological Bulletin. 114: 376-390. Messick DM, Sentis KP. 1985. Estimating social and nonsocial utility functions from ordinal data. European Journal of Social Psychology. 15: 389-399. Patterson BR, Bettini, LM 1993. Age, depression, and friendship, Development of a general friendship inventory. Communication Research Reports, 10: 161-170. Perugini M, Gallucci M. 1998. A game-theoretical model of the internalized norm of reciprocity. Paper presented at the ESRC conference on Decision-making in theory and practice, stream “Applications of Game Theory”, Oxford, 2 July. Perugini M, Gallucci M. 2001. Differences between equality, fairness and reciprocity: The role of valence. Unpublished manuscript. Perugini M, Gallucci M, Presaghi F, Ercolani AP. 2001. The personal norm of reciprocity. Unpublished manuscript. Sattler DN, Kerr NL. 1991. Might versus morality explored: Motivational and cognitive basis for social motives. Journal of Personality and Social Psychology. 60: 755-765. Schwartz, SH.1977. Normative influence on altruism. In L Berkowitz (Ed.), Advances in Experimental Social Psychology (Vol. 10, pp. 222-275). New York: Academic Press. Ten Berge MA, De Raad B. 1999. Taxonomies of situations from a trait psychological perspectives: A review. European Journal of Personality. 13: 337-360. Ten Berge MA, De Raad B. in press. The construction of a joint taxonomy of traits and situations. European Journal of Personality. Trivers RL. 1971. The evolution of reciprocal altruism. Quarterly Review of Biology. 46: 35-57. Van Lange, PAM, Liebrand, WBG. 1991. Social value orientation and intelligence: A test of the goal-prescribes-rationality principle. European Journal of Social Psychology. 21: 273-292. Van Lange, PAM, Otten, W, De Bruin, ENM, Joireman, JA 1997. Development of RECIPROCATORS AND PROSOCIALS 28 FOOTNOTES prosocial, individualistic, and competitive orientations: Theory and preliminary evidence. Journal of Personality and Social Psychology. 73: 733-746. Van Lange, PAM. 1999. The pursuit of joint outcomes an equality in outcomes: An integrative model of social value orientation. Journal of Personality and Social Psychology, 77: 337-349. Wright JC, Mischel W. 1987. A conditional approach to dispositional constructs: The local predictability of social behavior. Journal of Personality and Social Psychology. 53: 1159-1177. 1 The agreement was calculated on 510 unique sentences after grouping together same or extremely similar sentences. 2 We would like to thank a referee for this suggestion. 3 The results of the analyses are virtually unchanged if individualists and competitors are grouped together as proself (Van Lange & Liebrand, 1991). 4 An ANCOVA considering the total score in the three scales of reciprocity produced a similar result, with the scores in reciprocity showing a significant interaction with the factor valence (F(1,104)=10.446, p=.002). This interaction effect was obviously due to the negative reciprocity scale. 5As a referee pointed out, the measure of the SVO and the dependent variable share some similarities. However, it is unclear how this could have produced such a consistent pattern of results. If any, this similarity could explain the main effect of SVO, but not the other RECIPROCATORS AND PROSOCIALS 29 significant effects nor the lack of interactions concerning SVO RECIPROCATORS AND PROSOCIALS 30 Table 1. Study 1: Cross-frequencies of [event] [behavior] valences for reciprocity, cooperativeness, and hostility. a) Reciprocity [Behavior] Negative Neutral Positive Negative 21 0 1 22 [Event] Neutral 0 3 6 9 Positive 2 2 140 144 23 5 147 175 b) Cooperativeness [Behavior] Negative Neutral Positive Negative 2 0 22 24 [Event] Neutral 3 3 123 129 Positive 0 2 30 32 5 5 175 185 c) Hostility [Behavior] Negative Neutral Positive Negative 49 0 0 49 [Event] Neutral 135 2 3 140 Positive 51 0 0 51 235 2 3 240 RECIPROCATORS AND PROSOCIALS 31 Table 2. Study 1: Conditional probabilities of [event] [behavior] valences for reciprocity, cooperativeness, and hostility. a) Reciprocity [Behavior] Negative Neutral Positive Negative .954 0 .046 [Event] Neutral 0 .333 .667 Positive .014 .014 .972 b) Cooperativeness [Behavior] Negative Neutral Positive Negative .083 0 .917 [Event] Neutral .023 .023 .954 Positive 0 .063 .937 c) Hostility [Behavior] Negative Neutral Positive Negative 1 0 0 [Event] Neutral .964 .014 .022 Positive 1 0 0 RECIPROCATORS AND PROSOCIALS 32 Table 3. Study 1: Action congruency for reciprocity, cooperativeness, and hostility (percentages). Disposition Action congruency Reciprocity Cooperativeness Hostility Yes 75.4 24.9 25.8 No 24.6 75.1 74.2 RECIPROCATORS AND PROSOCIALS 33 Figure Captions Figure 1. Study 2: Interaction between valence and type of relationship RECIPROCATORS AND PROSOCIALS 34 5060708090100Pot. PartnerFriendType of relationshipAllocated payoffNegativePositiveValence

Tidak ada komentar:

Posting Komentar