Internal motives and external rewards
Pro-socials fight the competition, pro-selfs exploit cooperation
Berno Dussel University of Amsterdam
Student ID 10417729 Word Count: 8.270
Abstract
Information are a valuable resource. Thus sharing information, as sharing any other type of valuable good, is a highly strategic and socially motivated process. How much and with whom information is shared, is determined not only by the characteristics (cooperative or competitive) of the environment an individual finds itself in, but also the individual’s social value set that regulates the behavioral response to the situation. In an inter-group information pooling task, pro-socials and pro-selfs had the options of keeping, sharing, or lying about information. Pro-socials, who cared more about group-level goals, adapted more strongly to competition by retaining less and sharing more important information with their own group – thus strengthening the in-group position –, while pro-selfs, who focused on individual goals, exploited mainly a cooperative environment to further their own individualistic goals.
(Abstract word count = 132)
Keywords: hidden profile, information sharing, social value orientation, groups, deception, inter-group competition
Introduction
Information management is common practice during the exchange of important information. Husbands lie to their wives about their whereabouts, companies hide unpopular production practices, and politicians share only specific pieces of information during
important negotiations. However, in most group decision-making situations, information sharing serves as a crucial first step to reach the right decision. Without proper provision of information, decision-makers might settle for biased or disadvantageous solutions (Nijstad & De Dreu, 2012; Stasser & Titus, 2003). To disentangle the mechanisms behind information management and deception in the context of information sharing, the present research investigates how information sharing processes are influenced by an ever-present underlying agenda: social motives.
The influence of so-called social value orientations (SVO; Messick & McClintock, 1968) has been studied thoroughly in settings closely related to information sharing like negotiation (De Dreu, Beersma, Stroebe, & Euwema, 2006), bargaining (Van Dijk, De Cremer, & Handgraaf, 2004), and most importantly the investment in public goods (Rand, Greene, & Nowak, 2012). Here, the individual has a certain amount of valuable resources at her disposal and is offered two basic choice options: invest in the public good and thus create a better outcome for the whole group (e.g. by paying taxes to build roads available to
everyone), or keep these investments to oneself and free ride on the potential investments of others (e.g. by not paying taxes but still benefitting from the roads). Extant work in social psychology and economics suggests that the decision to cooperate and to invest (or not) partly depends on individual differences in social value orientation: Pro-socials strive for high joint outcomes and fairness and thus tend to choose the first option, sacrificing their resources to benefit the group. Pro-selfs, on the other hand, strive to maximize their own outcomes and
thus tend to choose to retain their resources and free-ride on the investments of others (Nijstad & De Dreu, 2012; Rand et al., 2012).
In an inter-group setting, these basic motivations can take on a radically different meaning. While pro-selfs stick to the original plan and retain their resources, pro-socials still show cooperation – but only towards their own in-group, not towards a (possibly competing) out-group, a phenomenon called parochial altruism (Baumgartner, Schiller, Rieskamp, Gianotti, & Knoch, 2013; Burton-Chellew & West, 2012; De Dreu et al., 2010). Being rooted deep in evolutionary adaptability, this bias to strengthen one’s in-group against outside threats can be observed in all kinds of groups, whether they are really competitors for some resource or just incidentally formed, like teams in school sports (c.f. social identity theory; Tajfel, 1974, 1982). For example, De Dreu (2010) classified people as pro-self or pro-social and asked them to invest (or not) in in-group love – benefitting other in-group members at a cost to themselves – and/or out-group hate – benefitting in-group members and
simultaneously hurting out-group members. Results showed that pro-selfs kept most of their endowment, and invested similarly low amounts in in-group love and out-group hate. In contrast, pro-socials invested substantially in in-group love and invested as much in out-group hate as pro-selfs. From this pattern, it was concluded that pro-socials are parochial co-operators rather than universal co-co-operators (in which case they should have invested less to nothing in out-group hate).
In the context of information sharing, research has largely focused on so-called hidden profile scenarios (e.g. Scholten, van Knippenberg, Nijstad, & De Dreu, 2007; Schulz-Hardt, Brodbeck, Mojzisch, Kerschreiter, & Frey, 2006; Schulz-Hardt & Mojzisch, 2012; Stasser & Titus, 1985, 2003). Here, group members each only possess partial information and depend on each other to share their unique information to be able to reach the best decision.
thus motivational processes have largely been neglected in this line of research (Steinel, Utz, & Koning, 2010). Taking a closer look, it might be reasonable to argue that – as in the investment scenarios described above – here, too, individuals possess valuable resources in the form of information which they can either share with others or keep for themselves. Thus social motives should drive tendencies to share, withhold, or even distort information (De Dreu & Boles, 1998; Steinel & De Dreu, 2004; Steinel et al., 2010; Toma & Butera, 2009).
A first study speaking to this possibility was conducted by Steinel et al. (2010). In their study, they gave a reward either to the whole group when any individual member gathered enough information in a modified hidden profile task (the cooperative incentive; c.f. Schulz & Pruitt, 1978) or gave the reward to every individual participant that had gathered enough information (the individualistic incentive). In addition, to shed even more light on the motivational processes behind information sharing, Steinel et al. (2010) also distinguished between information that was important for solving the task, and information that was rather unimportant. They found that, while participants in the cooperative condition shared their valuable information freely, participants in the individualistic condition tended to keep important information to themselves, shared only useless information, and even distorted information (i.e. lied about the content of their information).
In a real-life business environment, however, individualistic and cooperative goals may coexist and sometimes even clash. For example, a common practice in sales teams is handing out a special bonus to the top sales person at the end of the quarter. At the same time, the company aims to achieve the best possible company-wide results and thus facilitates cooperation among team members. Thus, in environments like these, not a singular social motive may be enhanced by externally set goals, but instead the individual is forced to decide which goal to pursue on the basis of his or her own set of values and adapt its behaviour accordingly.
The present research is aimed to shed light on the differences between pro-socials’ and pro-selfs’ behaviour in these situations, where cooperative information pooling is critical but external goals might facilitate or discourage cooperation. I argue that sharing information will follow similar behavioural patterns as investing material goods not only on the individual level but also on the group level. Pro-selfs will attempt to maximize own outcomes and keep important information to themselves, while pro-socials will share information, but only parochially with their in-group.
Building on the findings of Steinel et al. (2010), motivational differences should be mainly found for private (i.e. unknown to other participants) but not public information, as the latter has no additional value that can be kept from or given to other participants. For private information however, differences should occur between sharing (and keeping) of important information (that is most valuable in the task) and unimportant information with less value. Thus, compared to unimportant information:
(H1) Overall, pro-socials will share more important information with the in-group than pro-selfs.
This behaviour, however, also depends on the environment the individual finds him- or herself in. In a cooperative inter-group environment (e.g. in an organization that does not constitute team goals, but only company-wide and individual goals), where group goals are relatively safe, individual goals will become more manifest, enhancing the natural tendency in pro-selfs to maximize their individual outcomes. Pro-socials, who lack this inherent tendency, on the other hand, should show behavioural changes to a much lesser amount. Thus, compared to unimportant information:
(H2a) In a cooperative inter-group environment, pro-selfs will take the opportunity to maximize individual outcomes by retaining more and sharing less important information with the in- and out-group.
(H2b) As the group reward is safe, pro-socials as well will focus more on the individual goal, retaining more and sharing less important information with the in- and especially the out-group. Lacking the tendency to be selfish, differences will be weaker for pro-socials than pro-selfs.
(H2c) Additionally, both pro-socials and pro-selfs may engage in deceptive behaviour to enhance their individual position. While pro-selfs will lie to all other parties, pro-socials will only lie to the out- but not the in-group.
In contrast, in a competitive inter-group structure (e.g. in a company that honours only the best-performing team as well as the best-performing employee with a special bonus), the competition of the out-group threatens the achievement of in-group goals, thus enhancing the motivation to protect one’s group reward against the competition. In such a scenario, both socials and selfs should will protect their own group against the out-group, but pro-socials will put all their resources in these protective measures, while pro-selfs will also stay focused on the individual reward. Thus, compared to unimportant information:
(H3a) In a competitive inter-group environment, pro-socials will strongly cooperate parochially by keeping less important information to themselves, sharing more important information with the in-group, and less important information with the out-group.
(H3b) This effect will be weaker for pro-selfs, who also try to achieve the individual goal. Thus, pro-selfs, the same as pro-socials, will refrain from sharing important information
with the out-group, but will at the same time keep more important information to themselves and share less important information with their group members.
(H3c) To fight competition, both pro-socials and pro-selfs may also try to deceive the out-group by sharing more distorted important information.
Finally, in an independent inter-group environment (e.g. in a company rewarding high team performance – but not only the best team – as well as individual performance with a bonus), out-group interaction is rather uninteresting, thus sharing as well as deception should be low. Concerning the in-group however, behaviour will be driven by social motives. Pro-socials will share more and keep less important information, trying to achieve the group reward, while pro-selfs will share less and keep more important information to achieve the individual reward. Thus, compared to unimportant information:
(H4a) In an independent inter-group environment, pro-socials will share more
important information with the in-group, less or no important information with the out-group and keep lower amounts of important information for themselves.
(H4b) Pro-selfs will keep more and share less important information with both the in- and out-group.
(H4c) Deception should be rather low in the independent setting as out-group aggression is not relevant and lying to the in-group might hurt the achievement of the group reward.
Method Participants
175 participants (45 male, mean age 23.15 years, SD = 6.18) were recruited from the student body of the University of Amsterdam through an online recruiting system. The study was approved by the Psychology Ethics Committee of the University of Amsterdam, and adhered to ethical guidelines set forth by the APA. Participants provided written informed consent, were allocated at random to one of three experimental conditions (see below) and tested individually using a computer-guided procedure. After completing the study, participants were rewarded with either 10€ or course credit.
Design
Participants were evaluated on their social value orientation using a decomposed game task, dividing them into pro-social and pro-self subjects. They then completed an information pooling task with – depending on condition – a cooperative, independent, or competitive inter-group structure. Information available to participants during the task differed in levels of importance and whether the information were private (i.e. only known to the individual) or public (i.e. known to everyone involved in the game). Thus the design took the form of a 2 (social value orientation: pro-social vs. pro-self) x 3 (task structure:
cooperative vs. independent vs. competitive) x 2 (item importance: low vs. high) x2 (item availability: public vs. private) mixed model design with social value orientation and task structure serving as between subjects factors and item importance and availability as within subjects factors. Participants’ behavior during the information pooling task served as the main dependent variable. Participants had five options to manage their available information: for each of twelve pieces of information, participants decided whether to (a) keep the
information with the out-group, (d) lie to (i.e. share distorted information with) the in-group, or (e) lie to the out-group.
Materials & Procedure
Social Value Orientation
Participants first completed a well-established decomposed game measure (Griesinger & Livingston, 1973; Kuhlman & Marshello, 1975; Van Lange, 1999) to assess their
respective social value orientation as either pro-social or pro-self. For this task, participants were asked to choose between different alternatives of distributing points between themselves and another person, thus showing a participant’s preference for pro-self or pro-social options. The task was introduced as follows: “You will see nine decisions in which you have to make a choice. Your choice influences the amount of points you and some other person will get. Think of the points as something that is valuable to you, to which you attach great
importance. The other person also attaches great importance to the points.” For example, subjects would choose between the following three options: (A) 570 to self and 300 to other; (B) 500 to self and 100 to other; (C) 500 to self and 500 to other. In this example, (A) represents the individualistic choice as own outcomes (570) are higher than in both of the other options (both 500). Option (B) corresponds to a competitive strategy as the advantage over the other person (500-100 = 400) is larger than in the other choice options (570-300 = 270 and 500-500 = 0, respectively). Finally, (C) represents the pro-social choice as, here, joint outcomes are the largest (500+500 = 1000) when compared to options (A) and (B) (570+300 = 870 and 500+100 = 600, respectively). To be reliably classified, subjects needed to choose consistently in at least six out of nine trials. 149 subjects were classified as pro-social (80) or pro-self (69; 58 individualistic and 11 competitive) in this way. 26 subjects failed to show a consistent pattern and were dropped from further analysis.
Following the decomposed game measure, participants completed a series of unrelated filler tasks to avoid a carry-over effect between the social value measurement and the information pooling task. Then, after participants were finished with the filler tasks, they engaged in the information pooling task adapted from Steinel et al. (2010).
Information pooling task: cover story
Participants read that, centuries ago, an old monk hid a treasure in the graveyard of his monastery. To protect the exact location of the treasure, he wrote seven letters, each
containing more or less important cues. In order to find the treasure, it would be necessary to combine the information from these letters. Now, centuries later, the monastery was bought by an investment company and one of the seven letters was discovered in the library. The investment company published this letter in a newspaper and asked people who might be in the possession of the remaining letters to come forward and in a combined effort find the treasure. By chance, the participant had acquired such a letter at a flea market. Participants read, that, after they had contacted the investment company, the company announced that all six letters were located and that due to regional separation the company had organized two separate meetings between three of the owners each. The participant would take part in one of these meetings.
Information pooling task: manipulation
Before the meeting, participants were informed about the reward the investment company had offered. Participants learned in the
(a) Cooperative condition: If anyone belonging to either one of the two groups gathered enough information, everyone in both groups would earn a general reward. In addition, the one to find the location of the treasure first would earn a bonus reward of the same amount.
(b) Independent condition: If anyone belonging to either one of the two groups gathered enough information, everyone in that individual’s group would earn a general reward. In addition, the one to find the location of the treasure first in his group would earn a bonus reward of the same amount. Note: in this condition everyone could earn the general reward and two bonus rewards could be earned. (c) Competitive condition: If anyone belonging to either one of the two groups
gathered enough information and no member of the other group had done so already, everyone in that individual’s group would earn a general reward. In addition, the one to find the location of the treasure first would earn a bonus reward of the same amount. Note: in this condition only members of one group could earn the general reward and only one bonus reward could be earned. Before going on, participants were asked several control questions to check whether they had understood the respective reward structure. Wrong answers allowed participants to read the instructions again.
Information pooling task: types of information
Participants then reviewed the information at their disposal during the meeting. Each participant received the same list of twelve pieces of information. This list was comprised of six cues from the published letter (i.e. public information known to everyone) and six cues from the participant’s own letter (i.e. private information known only to the individual). Furthermore, information were categorized as being either important (i.e. eliminated large areas of the compound from search) or unimportant for the treasure hunt (i.e. eliminated small areas), resulting in three pieces of information for every combination of
public/important, public/unimportant, private/important, and private/unimportant. Participants were informed about the meaning of these four different labels and again answered a set of control questions before being able to proceed to the main task.
Information pooling task: dependent measures
Participants’ task and the main dependent variable was to decide on how to proceed with their twelve pieces of information. Participants were informed that, for every piece of information, they could
(a) Keep the information to themselves
(b) Share the information with their own group (i.e. make the information public for everyone in the in-group)
(c) Share the information with the other group (i.e. make the information public for everyone in the out-group)
(d) Lie about the information to their own group (i.e. share a distorted version of the information)
(e) Lie about the information to the other group (i.e. share a distorted version of the information).
Participants could, for every piece of information, choose whether they wanted to share it and, if yes, with whom (in-group or out-group) and in which form (truth or lie). Self-evidently keeping and sharing information as well as sharing and lying were mutually exclusive for a single item.
After deciding on each piece of information, participants filled out a short
questionnaire containing several manipulation check items (see table 1, items were measured on a 7-point Likert scale ranging from 1 = “do not agree at all” to 7 = “totally agree”) and questions on socio-demographical data.
Results
Prior to analysis, I checked for significant differences between participants
game measure. No significant differences on the manipulation checks as well as information pooling emerged (all p > .09). Thus I felt confident to merge individualists and competitors into the group of pro-selfs.
Furthermore, I checked for potential effects of gender and field of study. Although an initial multivariate analysis of variance (MANOVA) revealed a few main effects for both variables, these effects were diminished or completely removed when controlling for social value orientation and, when controlled for, did not change the results of the calculations below. They were thus disregarded for the following analyses.
Manipulation checks
To confirm manipulations, in a first step I conducted a varimax rotated principal component analysis (PCA) on the items of the final questionnaire. Table 2 shows item loadings on the seven factors that were generated using the Kaiser criterion (Eigenvalue > 1). According to the characteristics of the items loading on the seven factors (see table 2), I interpreted the factors as follows: in-group love (items 5, 7, 10, 11, 18, 25, 26, 33, 36, and 38), motivation to deceive (items 2, 19, 20, 21, and 32), motivation to share (items 1, 3, 4, 5, 18, and 31), motivation to universally cooperate (items 12, 16, 27, 28, 37, and 39), selfishness and distrust (items 8, 13, 14, 15, 16, 17, 29, and 30), fear (items 22, 23, 24, and 25), and competitiveness (items 8, 9, 10, 15, 34, and 35). For reference, item wordings are listed in table 1.
In a second step, I then conducted a multivariate analysis of variance (MANOVA) with SVO and task structure as independent variables and the newly generated factor scores as dependent variables.
Social value orientation: As expected, pro-socials (M = .16, SD = .11) showed a higher motivation than pro-selfs (M = -.20, SD = .12) to take care of their in-group (F(1, 147) = 4.86, p = .03) and to cooperate with both groups universally (M = .12, SD = .12 vs. M =
-.16, SD = .11, F(1, 147) = 3.54, p = .06). On the other hand, selfs, compared to pro-socials, were more selfish and distrustful of other participants (M = .19, SD = .12 vs M = -.16, SD = .11, F(1, 147) = 4.41, p = .04), showed higher levels of fear (M = .21, SD = .12 vs. M = -.18, SD = .11, F(1, 147) = 5.73, p = .02), and were somewhat more competitive (M = .17, SD = .12 vs. M = -.14, SD = .11, F(1, 147) = 3.51, p = .06). Congruently, pro-selfs (M = .13, SD = .12) showed a somewhat higher motivation to make use of deception than did pro-socials (M = -.14, SD = .11 F(1, 147) = 2.78, p = .10), while there was no difference in their basic motivation to share information (F(1, 147) = 1.34, p = .25).
Task structure: Interestingly, while the motivation to universally cooperate was – as expected – highest in the cooperative setting (M = .44, SD = .14), lower in the independent setting (M = -.07, SD = .13), and lowest in the competitive setting (M = -.44, SD =.13, F(2, 146) = 11.24, p < .01), ingroup love was significantly lower in the cooperative setting (M = -.35, SD = .14) when compared to the independent (M = .10, SD = .14) and the competitive (M = .20, SD = .14) settings (F(2, 146) = 4.35, p = .02). This may be due to the fact that
participants were much more afraid of other participants being selfish and trying to free-ride in the cooperative (M = .33, SD = .14) than in both the independent (M = -.25, SD = .14) and the competitive (M = -.04, SD = .14) settings. While participants in the competitive condition indicated a higher motivation to share information (M = .27, SD = .14) than in the cooperative (M = -.18, SD = .15) and independent (M = -.17, SD = .14) settings (F(2, 146) = 3.42, p = .02), they did not show any differences in their motivation to lie and neither their selfish nor their competitive motivation (all F < .59, all p > .56).
The interaction effect between SVO and task structure did not qualify these results further, with only a marginal effect on the fifth factor (selfishness and distrust, F(2, 143) = 2.81, p = .06) and no effects on all other factors (all F < 1.61, all p > .20).
Information pooling
Provision of information was scored by counting the amount of cues from each of the four categories (public/important, public/unimportant, private/important, and
private/unimportant) for each of the five behavioral possibilities (keep, share with in-group, share with out-group, lie to in-group, and lie to out-group). As each category had three items, scores ranged from zero to three.
Overall, participants shared an average of 6.3 pieces of information with their in-group, shared 3.4 pieces of information with the out-in-group, and kept 2.3 pieces of
information to themselves. Participants lied on average about 1 piece of information to their in-group and about 1.7 pieces of information to the out-group (figure 1).
As effects were mainly predicted for private information, but not publicly available and thus rather worthless information (an effect seen in previous studies using the same methodology, c.f. Steinel et al., 2010), I ran two sets of MANOVA, two each for private and public information, each with SVO and task structure as between-subjects factors and
repeated measures on information importance (important vs. unimportant). Differences in the amount of important (in contrast to unimportant) information kept, shared with the in-group, and shared with the out-group served as dependent variables in the first set of MANOVA to analyze differences in sharing behavior. In the second pair of MANOVA, deceptive behavior towards the in- and out-group served as dependent variables.
Information sharing: Sharing of public information differed solely for the factor importance (F(3, 141) = 5.54, p < .01). Participants kept more important than unimportant information (F(1, 147) = 4.72, p = .03), shared similar amounts of both types of information with the in-group (F(1, 147) = .34, p = .56), and shared less important than unimportant information with the out-group (F(1, 147) = 13.01, p < .01). There were no significant
differences for SVO, task structure, or any of the possible interaction terms for public information (all F < 1.52, all p > .21).
For private information, however, patterns were much more complex (tables 3 and 4). Sharing differed again for importance (F(3, 141) = 26.27, p < .01), with participants keeping significantly more (F (1, 147) = 35,74, p < .01) and sharing significantly less important than unimportant information with both the in-group (F(1, 147) = 22.25, p < .01) and the out-group (F(1, 147) = 49.75, p < .01). In addition, there was a main effect for task structure (F(6, 282) = 3.76, p < .01), with participants keeping similar amounts to themselves (F(2, 143) = .23, p = .80) and sharing similar amounts with the in-group (F(2, 143) = .38, p = .69) across conditions, but sharing significantly less information with the out-group in the competitive condition than in the other two conditions (F(2, 143) = 9.19, p < .01). While there were no effects for SVO (F(3, 141) = .61, p = .61) and the interaction term of SVO and task structure (F(6, 282) = .69, p = .66), all possible interactions with importance reached high significance. The basic main effect of importance was thus further qualified by significant interaction effects between both importance and task structure (F(6, 282) = 5.78, p < .01) and importance and SVO (F(3, 141) = 4.54, p < .01), and a triple interaction effect between importance, task structure and SVO (F(6, 282) = 3.11, p < .01).
The interaction between importance and SVO revealed that pro-socials and pro-selfs did not differ in keeping information (both kept higher amounts of important than
unimportant information F(1, 145) = .64, p = .43), nor sharing information with the out-group (both shared higher amounts of unimportant than important information, F(1, 145) = .94, p = .34), but did differ in the amounts of information shared with the in-group. While pro-selfs shared significantly less important than unimportant information, pro-socials shared similar amounts of both types of information (F(1, 145) = 10.81, p < .01).
The interaction between importance and task structure further revealed that only in the competitive condition, participants kept and shared with the in-group similar amounts of important and unimportant information, while they kept more (F(2, 143) = 3.04, p = .05) and shared less (F(2, 143) = 6.52, p < .01) important than unimportant information in both the cooperative and independent conditions. With regard to the out-group, participants shared significantly lower amounts of important rather than unimportant information in both the cooperative and competitive conditions (although overall sharing was significantly lower in the competitive setting, see above), but shared similar amounts of both types in the
independent condition (F(2, 143) = 10.44, p < .01).
Finally, the triple interaction between importance, SVO, and task structure specified these results even more, showing that, in a cooperative setting, pro-selfs kept more important than unimportant information to themselves, and, although sharing high amounts of
unimportant information, shared nearly no important information with both the in- and out-group (figure 2b). Pro-socials showed similar effects, but to a much lesser extent, especially for sharing with the in-group where pro-socials shared only marginally less important than unimportant information with their team mates (figure 2a). In the independent setting, with only symbolic competition present, pro-selfs kept lower amounts of important information than in the cooperative condition, negating the difference between important and unimportant information. They also shared less unimportant information with the in-group, diminishing the difference between both types of information, and less unimportant information with the out-group, totally negating the difference between important and unimportant information (figure 2d). Pro-socials kept similar amounts as in the cooperative setting (i.e. kept
substantially more important than unimportant information), and shared again somewhat less important than unimportant information. With regard to the out-group, though, pro-socials shared less unimportant information, bringing the amount shared to a similarly low level as
important information (figure 2c). In the condition featuring real competition, pro-selfs did adapt their behavior directed towards the out-group and shared nearly no important nor unimportant information with the out-group. However, within their own group, they still kept higher amounts of important than unimportant information to themselves and also shared somewhat less important than unimportant information with the in-group (figure 2f). Pro-socials, on the other hand, showed a much stronger reaction to competition, keeping only very low amounts of both important and unimportant information, sharing significantly lower amounts of important than unimportant information with the out-group, and, most
importantly, reversed their behavior, now sharing somewhat more important than unimportant information with their team mates (figure 2e).
Deception: Similarly to above, use of deceptive behavior did not substantially differ for public information. The conducted MANOVA revealed no main effects for importance, task structure, or SVO (all F < 1.20, all p > .31), only marginal effects for the interactions between SVO and task structure (F(4, 284) = 2.27, p = .06) as well as between SVO and importance (F(2, 142) = 2.74, p = .07), and no effects on the other possible interactions (both F < .24, both p > .92).
Again, results for private information were somewhat more complex (table 5). While there were no main effect for importance (F(2, 142) = 1.85, p = .16) nor task structure (F(4, 284) = .97, p = .39), there was a main effect for SVO (F(2, 142) = 5.28, < .01), showing that pro-selfs lied more to the out-group than pro-socials (F(1, 147) = 8.5, p < .01), but the two groups did not differ for in deceiving the in-group (F(1, 147) = 2.13, p = .15). There were no simple interaction effect involving task structure (both F < 1.54, both p > .19), but an
interaction effect between importance and SVO (F(2, 142) = 4.02, p =.02) qualified the above main effect further and revealed that both pro-socials and pro-selfs lied more about important than unimportant information to the out-group, but regarding the in-group pros-selfs lied
more while pro-socials lied less about important than unimportant information (F(1, 145) = 7.32, P < .01). Finally, the significant triple interaction between importance, SVO, and task structure (F(4, 284) = 3.86, p < .01) qualified the results even further. It revealed that the above effect was solely due to the cooperative setting, as neither pro-selfs nor pro-socials differed substantially in the amount of lies about important or unimportant information in the independent and competitive settings (figure 3a-3d). In the cooperative setting, however, pro-selfs showed similar (and rather high) amounts of deception about important and unimportant information towards the out-group and lied also significantly more about important
information, and about nearly no unimportant information to the in-group (figure 3f). Pro-socials, on the other hand, lied more about important than unimportant information to the out-group, but lied, compared to unimportant information, about next to no important information to their own group (figure 3e; F(2, 143) = 4.50, p < .01).
Discussion
The results discussed above demonstrate the nature of information pooling as an intrinsically strategic process. It does not surprise that differences in sharing behavior occurred predominantly for private information, the type of information that was valuable in our scenario. Differences for publicly known information, which was essentially worthless to the group goal, on the other hand, were very scarce.
Results show that, overall, while pro-socials and pro-selfs do not differ in the amounts of information kept, pro-socials do indeed share more important than unimportant
information with the in-group, while pro-selfs share mostly unimportant and lower amounts of important information with their team mates (hypothesis 1). In an environment that features cooperation and no threat for the group reward, pro-selfs focus on achieving their individual rewards, keeping more and sharing less (with both in- and out-group) important
rather than unimportant information (hypothesis 2a). Interestingly, pro-selfs also show signs of impression management, as they share very high amounts of unimportant information to signal cooperation and potentially elicit reciprocity. Pro-socials as well show this tendency but to a much lesser extent, especially with regard to information shared with the in-group (hypothesis 2b). On the other hand as predicted by hypothesis 3a, facing strong competition, pro-socials indeed adapted to the situation by being less selfish (i.e. they kept only low amounts of both important and unimportant information), showing more cooperation towards their own group (i.e. the only condition, where participants actually shared more important than unimportant information), and shutting out the out-group (i.e. by sharing next to no important information). Pro-selfs also adapted to the situation, but to a much lesser amount, as they only regulated their cooperation with the out-group (i.e. did not share any
information), but did not substantially decrease the amount of information retained nor increase cooperation with the in-group (hypothesis 3b). Finally, in a setting featuring a reward system that is independent of out-group actions, pro-socials and pro-selfs showed similar behavior. Although their behavioral patterns slightly differed, the actual amounts of important and unimportant information kept and shared did not differ substantially between pro-socials and pro-selfs. Both seemed to act on the individual goal, keeping more important than unimportant information to themselves, while sharing less important than unimportant information with the in-group, and similarly low amounts of both information types with the out-group. Thus, while hypothesis 4b (that pro-selfs would focus on the individual goal) was supported by the results, hypothesis 4a (that pro-socials would focus on the group goal) was not. I will discuss potential reasons for this effect below.
These results were further qualified by the option to deceive. Here, results show, that deception was used rather rarely in both the competitive and independent settings, but that overall pro-selfs deceive more than pro-socials. Important differences occurred in the
cooperative condition, however: in addition to the low amount of cooperation pro-selfs showed (see above), they also engaged rather strongly in deceptive behavior and did not differentiate between the out- and in-group. Pro-socials, on the other hand, engaged in deception as well to potentially strengthen their own chance of earning the individual reward, but did only lie about important information to the out-group, and next to never to the in-group. Thus, hypotheses 2c and 4c were supported by the results, while hypothesis 3c was not.
Theoretical and practical implications
The present study replicates findings from and adds to two different but related bodies of research. First of all, classical hidden profile research (Scholten et al., 2007; Schulz-Hardt et al., 2006; Schulz-Hardt & Mojzisch, 2012; Stasser & Titus, 1985, 2003) regularly finds that participants prefer to talk about information that is already known to everyone and shy away from sharing their own information. My findings replicate this effect, but allow further qualification in line with Steinel et al. (2010) via the differentiation between important and unimportant information. Results point to the conclusion that while there is a preference to withhold private information over public information, this effect is based mainly on important private information, but not on private information that is deemed unimportant. These results showcase the process of information sharing as a motivated process, where publicly known and unimportant information are more or less worthless and thus freely shared and more rarely retained, while important information that is unknown to other participants is strategically shared or withheld, depending on the environment and individual social motivation.
Second, results add to the knowledge base of social decision making. The present study shows that sharing information is indeed a special variant of investing resources – even though information, in sharp contrast to monetary investments, are not lost when shared. The
present results transfer previous findings in the field of social decision making and behavioral economics to the setting of information processing, showcasing that pro-selfs strategically withhold and even distort their information to potentially gain an individual edge, especially when confronted with an exploitable environment. Pro-socials, on the other hand, are more concerned with the well-being of their in-group and thus adapt more strongly to out-group threats and shy away from actively harming the outcomes of their team mates.
Additionally, results showcase that, although potential benefits could be reaped in the present scenarios from actively harming the out-group to strengthen the in-group in a truly competitive environment, parochialism in pro-socials takes mostly the form of in-group love (i.e. cooperating with one’s own group), but not group hate (i.e. actively harming the out-group). In contrast, attacks on the out-group mostly occur in a cooperative environment, when pro-selfs see the chance to harm out-group (and in-group) members to gain a personal edge.
Strengths and limitations
First, using a student-only sample could skew results and impair the transfer to a more general sample. This criticism is of course true for most studies in social psychology that are conducted in a higher learning environment. Generalizability with regard to social value orientations has up to date not been a problem, though.
Second, participants in the study could earn two rewards, a general group reward and an individual reward. Assuming at least one person would gather enough information, in the cooperative condition, the group reward could be earned with a chance of 6/6 (100%) and the individual reward with a chance of 1/6. In the competitive condition, the group reward could be earned with a chance of 3/6 (50%) and the individual reward with a chance of 1/6 as well. Finally, in the independent condition, the group reward could be earned with a chance of 3/3 (100%), while the individual reward could be earned with a chance of 1/3. From these
chances, it is apparent that only in the competitive condition, the group reward was
endangered, potentially facilitating cooperation. However, chances for the individual reward also changed across conditions, making it potentially easier to earn the individual reward in the independent condition compared to the cooperative and competitive conditions. As I did not correct for these different chances (e.g. by offering a higher individual reward in the cooperative and competitive settings), it is very well possible that the rather selfish behavior of pro-socials in the independent setting (contrary to my hypotheses) was due to a higher chance of earning the individual reward, thus making it much more profitable to be selfish. Future research could potentially verify this idea by for example offering only half the group reward as individual reward in the independent setting.
Third and most importantly, it is important to note, as reviewers of this paper did, that social value orientation is often manipulated using external reward systems (c.f. Schulz & Pruitt, 1978; Steinel et al., 2010), aiming to overwrite intrinsic motivations by external motivators. The present study manipulates external rewards but also takes into account social values as an independent factor, potentially leading to confounding variables. However, this approach would expect only (potentially similar) main effects for both task structure and SVO, and no interaction. Results clearly indicate that, while condition generally shifted cooperation in the respective direction, pro-socials and pro-selfs reacted very differently to the manipulation, shown by the interaction effects between task structure and SVO. This interaction cannot be explained by the confound, but only by the original hypothesis that an external reward stimulates adaption of intrinsic motivations, but does not shift them in the same way for pro-selfs and pro-socials. For future research, the manipulation of SVO via reward systems should thus be done carefully, to take into account different patterns of environmental adaptations for pro-socials and pro-selfs.
Conclusion
The present study showcases that in the setting of social dilemmas, information are indeed a special form of currency. Important and valuable information are traded strategically and with great care, while less valuable information are shared more freely or used to signal cooperation, when in reality no real cooperative intention exists. Most importantly, the study at hand reveals that pro-socials react stronger to a competitive environment and try to strengthen their in-group via increased cooperation as much as possible. Pro-selfs, on the other hand, react only to a lesser amount to competition and instead keep focused on their own, individual goals. Thus, especially in a very cooperative environment, where general rewards are not threatened, pro-selfs show only very low cooperation and may even resort to somewhat hostile behavior, deceiving the out-group as well as their own team mates. This hostility may, however, not always be noticeable as pro-selfs also engage in impression management, sharing a lot of less valuable information to mask their uncooperative behavior.
Results furthermore pose the question how profitable different reward structures really are. Cooperative environments may facilitate too much focus on individual goals, while competitive team reward systems are nurturing hostility between teams. Future research may be faced with the question how to motivate both pro-socially and individualistically
motivated individuals to engage in cooperative behavior, especially across team borders.
Table 1
Manipulation check items used in the final questionnaire Item No. Wording (translated from Dutch)
01 I shared a lot of information. 02 I distorted a lot of information. 03 I shared a lot of private information. 04 I shared a lot of important information.
05 I shared a lot of important information to my group.
06 I feel responsible for the outcomes of the other five participants.
07 My foremost goal is that my group members and I receive the general reward. 08 My foremost goal is that I receive the bonus reward.
09 I think it more important that everyone receives the general reward than that I receive the bonus reward.
10 I think it more important that everyone in my own group receives the general reward than that I receive the bonus reward.
11 I shared information to help my own group. 12 I shared information to help both groups.
13 I only shared information to get information back from the others. 14 I wanted more information than the other participants.
15 I wanted better information than the other participants. 16 I saw the members of the other group more as competitors. 17 I saw the members of my own group more as competitors.
18 It was important for me that all the members of my own group had enough information.
19 I distorted information to get an advantage over the other participants. 20 I distorted information to protect my private information.
21 I distorted information to be the first to find the treasure.
22 I feared that I might be the only one to share important information. 23 I feared that I might be the only one to share private information.
24 I feared that the other participants might prefer to keep their information to themselves.
25 I trusted that the members of my own group would share information. 26 I trusted that the information from my own group would not be distorted. 27 I trusted that the members of the other group would share information. 28 I trusted that the information from the other group would not be distorted. 29 I did not trust the other participants.
30 I did not like to share information from my own letter. 31 You do not lose the information, thus you can freely share. 32 I wanted to share my own information distorted.
33 I wanted that someone from my own group (not necessarily myself) finds the treasure.
34 I wanted to find the treasure myself.
35 I wanted to be the first in my group to find the treasure. 36 I want to help my own group.
37 I want to help both groups.
38 I feel responsible for my own group. 39 I feel responsible for both groups.
Table 2
Manipulation checks: underlying factors and item loadings Item No. Factor 1: In-group love 2: Motivation to deceive 3: Motivation to share 4: Universal cooperation 5: Selfishness and distrust 6: Fear 7: Competition 36 .833 .132 -.214 .122 26 .726 .229 .119 -.225 33 .703 .206 -.116 -.265 38 .696 .283 -.203 .113 -.156 11 .664 -.134 .437 -.156 -.184 25 .623 .321 .212 -.400 .108 07 .605 -.116 .383 -.338 10 .581 -.198 .128 -.121 -.533 19 .902 .152 20 .871 .128 21 -.139 .859 -.196 .197 .163 02 .823 -.203 .107 32 .744 .129 .114 .244 03 .254 -.170 .772 .152 -.130 04 .326 .752 .102 -.160 -.147 -.160 01 .310 -.276 .689 .226 -.171 -.144 05 .528 -.103 .644 -.122 -.132 -.224 18 .536 .586 -.207 -.286 31 .227 -.153 .586 .287 -.209 06 .333 -.175 .508 .195 -.211 .169 37 .880 -.116 12 .843 -.115 39 .122 .821 .131 -.217 28 .194 .751 .200 -.162 .160 27 .124 .177 .720 .161 .213 13 -.105 .112 .748 .213 14 -.215 .223 -.245 .109 .676 .209 .280 17 -.348 .259 -.297 .620 .266 15 .283 -.255 .490 .243 .462 29 -.370 .246 -.309 .486 .339 .101 30 -.307 .395 -.265 .468 .276 .228 16 .279 .133 -.461 .464 .318 08 -.228 .270 -.353 .439 .103 .424 23 .183 .884 .130 22 .128 -.102 .128 .862 .245 24 -.176 .140 .346 .791 34 .252 -.201 .264 .288 .709 35 -.128 .254 -.191 .276 .316 .644 09 .354 -.316 .206 -.578
Table 3
Sharing behavior: Interaction effects
Measure F p Importance Multivariate 26.27 < .01** KEEP 35.74 < .01** SIG 22.25 < .01** SOG 49.75 < .01** * Condition Multivariate 5.78 < .01** KEEP 3.04 .05* SIG 6.52 < .01** SOG 10.44 < .01** * SVO Multivariate 4.54 < .01** KEEP .64 .42 SIG 10.81 < .01** SOG .94 .34 * Condition * SVO Multivariate 3.11 < .01** KEEP 4.09 .02* SIG 2.20 .12 SOG 5.29 < .01**
Note: significance p values are marked by stars: marginal significance at .10 level (*), significance at .05 level *;
Table 4
Sharing behavior: Means and Standard Deviations
Measure Task structure SVO Mean (SD) Mean Difference
important unimportant
KEEP cooperative prosocial 1.07 (1.21) .52 (.77) .55* proself 1.39 (1.01) .22 (.73) 1.17** independent prosocial 1.38 (1.14) .13 (.34) 1.25** proself .92 (1.22) .40 (.87) .52 competitive prosocial .64 (.81) .68 (1.07) -.04
proself 1.08 (1.32) .46 (.95) .62* SIG cooperative prosocial 1.10 (1.22) 1.61 (1.23) -.52(*)
proself .50 (.86) 2.39 (1.04) -1.89** independent prosocial 1.17 (1.17) 1.83 (1.13) -.67(*)
proself .72 (.98) 1.48 (1.26) -.76* competitive prosocial 1.84 (.14) 1.24 (1.30) .60(*) proself 1.04 (1.31) 1.65 (1.20) -.62(*) SOG cooperative prosocial .48 (.96) 1.26 (1.26) -.77** proself .06 (.24) 1.89 (1.23) -1.83** independent prosocial .58 (.97) .96 (1.12) -.38
proself .48 (.96) .76 (1.09) -.28 competitive prosocial .08 (.28) .68 (1.03) -.60** proself .12 (.43) .31 (.68) -.19 Note: significance levels for individual t-tests are symbolized by stars: marginal significance at .10 level (*),
Table 5
Deception: Means and Standard Deviations
Measure Task structure SVO Mean (SD) Mean Difference
important unimportant
LIG cooperative prosocial .13 (.34) .45 (.96) -.32(*) proself .67 (1.03) .00 (.00) .67** independent prosocial .21 (.51) .21 (.51) .00 proself .40 (.76) .48 (1.01) -.08 competitive prosocial .08 (.28) .24 (.66) -.16 proself .31 (.68) .19 (.40) .12 LOG cooperative prosocial .74 (1.13) .19 (.60) .55** proself 1.06 (1.16) .50 (.92) .56 independent prosocial .13 (.34) .38 (.92) -.25
proself .68 (1.03) .44 (1.00) -.24 competitive prosocial .52 (.92) .20 (.50) .32
proself .62 (1.02) .85 (1.12) -.23 Note: significance levels for individual t-tests symbolized by stars: marginal significance at .10 level (*),
Figure 1 Overall sharing
Note: KEEP = keeping information, SIG = sharing with in-group, SOG = sharing with out-group, LIG = lying to in-group, LOG = lying to out-group
.0 .2 .4 .6 .8 1.0 1.2 1.4 1.6 1.8 2.0
KEEP SIG SOG LIG LOG
av er ag e am ou nt of information type of behavior
Figure 2
Sharing patterns (private information only) for pro-socials and pro-selfs
Note: KEEP = keeping information, SIG = sharing with in-group, SOG = sharing with out-group, significance levels for pairwise comparisons (important vs unimportant) are symbolized by stars: marginal significance at .10 level (*), significance at .05 level *; and strong significance at .01 level **
* (*) ** 0.0 0.5 1.0 1.5 2.0 2.5
KEEP SIG SOG
Am ount of information Type of behavior
(a) Cooperative (pro‐social)
Important Unimportant ** ** ** 0.0 0.5 1.0 1.5 2.0 2.5KEEP SIG SOG
Am ount of information Type of behavior
(b) Cooperative (pro‐self)
Important Unimportant ** (*) 0.0 0.5 1.0 1.5 2.0 2.5KEEP SIG SOG
Amo unt of information Type of behavior
(c) Independent (pro‐social)
Important Unimportant * 0.0 0.5 1.0 1.5 2.0 2.5KEEP SIG SOG
Amo unt of information Type of behavior
(d) Independent (pro‐self)
Important Unimportant (*) ** 0.0 0.5 1.0 1.5 2.0 2.5KEEP SIG SOG
Am o unt of information Type of behavior
(e) Competitive (pro‐social)
Important Unimportant * (*) 0.0 0.5 1.0 1.5 2.0 2.5KEEP SIG SOG
Am o unt of information Type of behavior
(f) Competitive (pro‐self)
Important UnimportantFigure 3
Deception patterns (private information only) for pro-socials and pro-selfs
Note: LIG = lying to in-group, LOG = lying to out-group, significance levels for pairwise comparisons (important vs unimportant) are symbolized by stars: marginal significance at .10 level (*), significance
at .05 level *; and strong significance at .01 level **
(*) ** 0.0 0.5 1.0 1.5 LIG LOG Am ount of information Type of behavior
(a) Cooperative (pro‐social)
Important Unimportant ** 0.0 0.5 1.0 1.5 LIG LOG Am ount of information Type of behavior(b) Cooperative (pro‐self)
Important Unimportant 0.0 0.5 1.0 1.5 LIG LOG Amo unt of information Type of behavior(c) Independent (pro‐social)
Important Unimportant 0.0 0.5 1.0 1.5 LIG LOG Amo unt of information Type of behavior(d) Independent (pro‐self)
Important Unimportant 0.0 0.5 1.0 1.5 LIG LOG Am o unt of information Type of behavior(e) Competitive (pro‐social)
Important Unimportant 0.0 0.5 1.0 1.5 LIG LOG Am o unt of information Type of behavior(f) Competitive (pro‐self)
Important UnimportantReferences
Baumgartner, T., Schiller, B., Rieskamp, J., Gianotti, L. R., & Knoch, D. (2013). Diminishing parochialism in intergroup conflict by disrupting the right temporo-parietal junction. Soc Cogn Affect Neurosci.
Burton-Chellew, M. N., & West, S. A. (2012). Pseudocompetition among groups increases human cooperation in a public-goods game. Animal Behaviour, 84(4), 947-952.
De Dreu, C. K. W. (2010). Social value orientation moderates ingroup love but not outgroup hate in competitive intergroup conflict. Group Processes & Intergroup Relations, 13(6), 701-713.
De Dreu, C. K. W., Beersma, B., Stroebe, K., & Euwema, M. C. (2006). Motivated information processing, strategic choice, and the quality of negotiated agreement. Journal of Personality and Social Psychology, 90(6), 927-943.
De Dreu, C. K. W., Greer, L. L., Handgraaf, M. J. J., Shalvi, S., Van Kleef, G. A., Baas, M., . . . Feith, S. W. W. (2010). The neuropeptide Oxytocin regulates parochial altruism in intergroup conflict among humans. Science, 328(5984), 1408-1411.
De Dreu, C. K. W., & Boles, T. L. (1998). Share and Share Alike or Winner Take All?: The Influence of Social Value Orientation upon Choice and Recall of Negotiation Heuristics. Organizational Behavior and Human Decision Processes, 76(3), 253-276.
Griesinger, D. W., & Livingston, J. W. (1973). Toward a model of interpersonal motivation in experimental games. Behavioral Science, 18(3), 173-188.
Kuhlman, D. M., & Marshello, A. (1975). Individual differences in the game motives of Own, Relative, and Joint gain. Journal of Research in Personality, 9(3), 240-251.
Messick, D. M., & McClintock, C. G. (1968). Motivational bases of choice in experimental games. Journal of Experimental Social Psychology, 4(1), 1-25.
Nijstad, B. A., & De Dreu, C. K. W. (2012). Motivated information processing in organizational teams: Progress, puzzles, and prospects. In A. P. Brief & B. M. Staw (Eds.), Research in Organizational Behavior: An Annual Series of Analytical Essays and Critical Reviews, Vol 32 (Vol. 32, pp. 87-111). New York: Elsevier.
Rand, D. G., Greene, J. D., & Nowak, M. A. (2012). Spontaneous giving and calculated greed. Nature, 489(7416), 427-430.
Scholten, L., van Knippenberg, D., Nijstad, B. A., & De Dreu, C. K. W. (2007). Motivated information processing and group decision-making: Effects of process accountability on information processing and decision quality. Journal of Experimental Social Psychology, 43(4), 539-552.
Schulz-Hardt, S., Brodbeck, F. C., Mojzisch, A., Kerschreiter, R., & Frey, D. (2006). Group decision making in hidden profile situations: Dissent as a facilitator for decision quality. Journal of Personality and Social Psychology, 91(6), 1080-1093.
Schulz-Hardt, S., & Mojzisch, A. (2012). How to achieve synergy in group decision making: Lessons to be learned from the hidden profile paradigm. European Review of Social Psychology, 23(1), 305-343.
Schulz, J. W., & Pruitt, D. G. (1978). The effects of mutual concern on joint welfare. Journal of Experimental Social Psychology, 14(5), 480-492.
Stasser, G., & Titus, W. (1985). Pooling of unshared information in group decision-making - Biased information sampling during discussion. Journal of Personality and Social Psychology, 48(6), 1467-1478.
Stasser, G., & Titus, W. (2003). Hidden profiles: A brief history. Psychological Inquiry, 14(3-4), 304-313.
Steinel, W., & De Dreu, C. K. W. (2004). Social motives and strategic
misrepresentation in social decision making. Journal of Personality and Social Psychology, 86(3), 419-434.
Steinel, W., Utz, S., & Koning, L. (2010). The good, the bad and the ugly thing to do when sharing information: Revealing, concealing and lying depend on social motivation, distribution and importance of information. Organizational Behavior and Human Decision Processes, 113(2), 85-96.
Tajfel, H. (1974). Social identity and intergroup behaviour. Social Science Information, 13(2), 65-93.
Tajfel, H. (1982). Social Psychology of Intergroup Relations. Annual Review of Psychology, 33(1), 1.
Toma, C., & Butera, F. (2009). Hidden Profiles and Concealed Information: Strategic Information Sharing and Use in Group Decision Making. Personality and Social Psychology Bulletin, 35(6), 793-806.
Van Dijk, E., De Cremer, D., & Handgraaf, M. J. J. (2004). Social value orientations and the strategic use of fairness in ultimatum bargaining. Journal of Experimental Social Psychology, 40(6), 697-707.
Van Lange, P. A. M. (1999). The pursuit of joint outcomes and equality in outcomes: An integrative model of social value orientation. Journal of Personality and Social
Author’s Note
This work is presented as a research master’s thesis at the University of Amsterdam in summer 2014.
The author herewith declares that the paper submitted is his own work and that he used only the sources disclosed here.
Author contact information: Berno Dussel, Humboldtstr. 34, 81543 Munich, Germany, berno.dussel@gmail.com, UvA ID: 10417729.
Date of submission: August 23rd, 2014. Total word count: 9.104.
