In practice, it allows exploring heterogeneous beliefs and responses to competition because downward deviations are rational only if subjects believe that all the groups are not equally competitive.
We also explore the role of group size under the hypothesis that the likelihood to coordinate and the salience of between-group conflict are higher for smaller groups. In this paper we contribute to the group competition literature in two different fronts. First, by exploring in detail the informational effects of individual and group performance on between-group competition.
We find that while the relative group standing in the previous round decreased current cooperation i. Second, by showing that these informational effects are structurally similar with or without additional material incentives to compete. The only difference is that the intercept is shifted upwards i. These results show a complex dynamic regarding the problem of between-group competition with intra-group cooperation, and opens further questions regarding rewards at the individual and group levels.
Informed consent was obtained from all the participants, who were recruited through email solicitations. Due to the low-risk nature of this experiment, the voluntary participation of subjects, and the use of a consent form signed by them, approval from the ethics committee was not requested. However, the protocol followed all ethical standars common in economic experiments. Each session lasted between 70 and 90 min. This payment is about 1. In each session, participants were divided into six groups to play a voluntary contributions mechanism VCM for twenty rounds.
We assigned players to groups of either 3, 4, or 5 people and remained fixed during the whole session. Incentives including group performance were introduced in the last ten rounds, although information on individual and social comparison were provided after each round for the entire session.
This information was known by all subjects. At the end of each round participants were reminded of their contribution, and were informed about their group's aggregate contribution, their round earnings, their individual ranking among the whole population in the session but not within their group 5 , their group ranking based on aggregate earnings , and the highest and lowest earnings of the round among all subjects in the session.
The 20 rounds were divided into two stages of ten rounds each. We will call the first stage pseudocompetition because payoffs are not directly affected by the group's ranking. In this stage, comprising rounds 1—10, we have a standard VCM payoffs function. In the case of a tie between two or more groups, their multipliers are averaged.
It prevents that contributing becomes increasingly attractive with higher expectations of group performance. A similar incentive for group performance is proposed in Reuben and Tyran , with multipliers that are used for punishing groups not able to rank first.
Under their conditions the game has two symmetric equilibria: null contribution and full contribution. As in Reuben and Tyran , in our experimental setting the between-group incentives lead to some asymmetric equilibria with different contribution levels across groups. They appear when a single subject cannot alter the group's ranking given her beliefs about the other subjects' actions.
Their current payoff 1. Group size is an additional source of variation in our experimental design. We hypothesize that smaller groups are more reactive to the cues of between-group competition and therefore that they will be closer to the NE under competition for two reasons: first, the intra-group coordination required to raise the group's relative performance might be, in theory, easier to achieve e. Each unit in the common fund generated the same value irrespective of the group size.
Instructions were given orally and in written form. The instructions were followed by two examples on the board. The protocol read to the participants is included in the online Supplementary Material. After the tenth round new instructions were given for the competition stage. The new instructions emphasized the ranking calculations and transformation of the payoffs. A post experimental survey was conducted to collect data on participants' competitiveness and trustworthiness, as well as basic demographics.
The questions used to measure competitiveness are shown in the online Supplementary Material. In the pseudocompetition stage the average contribution was 3. We observe see Figure 1 the decay typical of a standard linear PG game Ledyard, ; Zelmer, , ranging from 4.
As expected from the theoretical predictions, contribution levels are higher once the between-group competition is materially incentivized.
In the competition stage the average contribution is 7. The increasing contribution rate, going from 6. Figure 1. Average contribution with and without group competition. The gray horizontal line represent the Nash equilibrium of the game in the pseudocompetition and the competition stages. We do not find an effect of group size, possibly because we primed the total value of each token in the common fund invariant with group size instead of priming the value of the MPCR which decreases with group size.
The NE outcome under competition, i. However, smaller groups were more likely to maximize their total contribution to the common fund 6. A shift in individualists' cooperative behavior, triggered by intergroup competition, is reported in Probst et al. Their test-based measure of individualism compared to collectivism is correlated with more contributions in a public goods game with intergroup dependent payoffs 7.
Figure 2. Total contribution per player with and without competition. In the left panel we plot the mean contribution per player under pseudocompetition vs. Marker size reflects the frequency of observations in each point of the grid. On the left panel of Figure 2 is observed that a small proportion of subjects 5. However, the statistical tests are not very likely to result significant given the low proportion of non-competitors.
We show in Figure 3 the effect of feedback on group and individual performance in subsequent contribution decisions. A comparison of the four panels in this figure reveals the following three patterns: i Better group performance reduces individual contributions, whereas better individual performance increases individual contributions. On the other hand, the nonlinearity in the response to individual performance becomes less pronounced with incentivized competition.
Figure 3. Changes in contribution with respect to relative performance. For treatments with 18 and 24 participants the individual ranking was rescaled to match the treatment with 30 participants. In C,D subjects in the worst performing group are excluded given their null payoff, leading to a large number of ties at the bottom of individual ranking. This method allows us to control for unobserved time invariant characteristics of the subjects.
We observe the negative and decreasing effect of group ranking as well as the positive and decreasing effect of individual ranking before and after introducing material incentives to compete. We also run the regressions excluding the participants belonging to the group that end up last in each round.
Our motivation for this robustness check is twofold: to check that the nonlinearities are not driven by subjects in groups that ranked last, who might be different from the other participants; and to exclude from the subsample the subjects with perfect correlation between individual and group ranking given their null payoff under competition. The nonlinear effect of group ranking is no longer significant under pseudocompetition , whereas the coefficients increase in magnitude under competition.
This result suggests that the incentivized inter-group competition induces more responsiveness to the social comparison, if compared to the pseudocompetition. The same pattern is observed across treatments: high individual performance and low group performance are followed by an increase in contributions.
The stark difference is a positive shift in the intercept once group competition is materially incentivized. Under competition the individual contribution is slightly reduced only in case of a high group ranking at least 5 and a low individual ranking at most 4. This reaction could be interpreted as a reciprocal response to the environmental cues suggesting higher contributions with respect to the other team members.
Figure 4. Predicted change in contribution as a function of group and individual performance. The left and right panels correspond to the predicted difference in contribution under pseudocompetition and competition using results from columns 1 and 3 in Table 2 , respectively. Red resp.
We perform three robustness checks to the regressions shown in Table 2. The results are reported in the online Supplementary Material. First, we show that all the coefficients are statistically different between treatments see Table B. Second, we compute again the regressions excluding the observations in which subjects were playing the NE, i.
Extending previous findings, we found sequential processing adjustments of the Simon effect as a function of the interdependency i. While sequential processing adjustments of the Simon effect in both the competition and cooperation condition were unaffected when alternating between responsible actors i.
Therefore, people keep their eyes on themselves when aiming at beating a co-actor and emerging as the winner.
For humans, it is nearly impossible not to interact with others Watzlawick et al. Beyond the significant role of exchanging information and communicating with each other, there are many instances in everyday life that require cooperation e.
Experimental approaches aiming at investigating the underlying mechanisms of social interactions in the laboratory typically use the Simon task Craft and Simon, In the standard, two-choice version of the Simon task, participants are asked to respond with a left or right keypress to a particular feature of the stimulus e. If the spatial location of the stimulus corresponds with the spatial location of the assigned response i. In contrast, if the spatial location of the stimulus and the assigned response differ i.
The difference between RTs or ERs in incompatible and compatible trials is called the Simon effect cf. Simon and Rudell, ; for an overview, see Simon, ; Lu and Proctor, This Stimulus—Response S—R compatibility effect is typically explained by the dimensional overlap model Kornblum et al.
Accordingly, the location of the stimulus is assumed to automatically activate a spatially corresponding response, which facilitates task performance on compatible trials and impairs performance on incompatible trials, because resolving the conflict between automatically activated and required responses takes time.
To investigate social interactions, Sebanz et al. In order to investigate the impact of performing the go—nogo task together with another person, Sebanz et al. While there was no significant S—R compatibility effect in the individual go—nogo condition, there was a compatibility effect when two participants performed the same go—nogo Simon task together, and this finding became known as the social or joint go—nogo Simon effect JSE; Sebanz et al.
Thus, the social task context seems to modulate the allocation of attention toward the specific S—R associations cf. Baess and Prinz, This explanation nicely fits with the existing theoretical frameworks aiming to explain the emergence of JSEs: The action co-representation Sebanz et al.
Accordingly, spatially assigned stimuli and responses of the whole task set are considered to be represented, which facilitates task performance in cases of an S—R match, but interferes with performance when there is an S—R mismatch. Consequentially, discriminating alternative action events should be more challenging the more similar those events are, resulting in varying effect sizes with varying degrees of similarity i.
These assumptions nicely converge with previous findings showing increased non-social and social JSEs with increasing similarity of alternative action events: e. Thus, while social variables appear to play an influential role for the co-representation account, the referential coding account highlights the role of the similarity between alternative action events irrespective of its non- social nature.
In both cases, however, attention allocation toward the spatially distinct alternative actions seem to impact the cognitive representation thereof and, what is more, subsequent behavior for an attentional focusing account of joint compatibility effects, see Dittrich et al. One straightforward approach to tackling this issue is to manipulate the relationship between, or the interdependence of, interacting individuals. While the offending behavior of an intimidating confederate reduced the JSE as compared to a friendly co-actor Hommel et al.
Using an auditory joint go—nogo Simon task i. In advance of the Simon task, participants were instructed that either: i the ten best performing subjects will win ten euros independent group , ii each actor of the five best performing teams will earn five euros cooperative group , or iii ten team winners will be randomly selected for the ten euros reward competition group. Results revealed a larger JSE in the dependent groups i.
This finding has been taken to suggest that interdependency leads to a stronger attentional focus on the partner, and therefore to stronger shared representations and to a larger JSE.
Iani et al. In sharp contrast to Ruys and Aarts , the results revealed a significant difference between dependent groups, with a JSE in the cooperative group but no JSE in the competitive group. Thus, considering the co-actor might selectively occur when aiming to beat others as a team, but the exact opposite takes place when having an opponent.
Further support for the crucial role of the exact type of interdependency is provided by a study of Ruissen and de Bruijn showing a smaller JSE in a group of participants that played Tetris in a competitive as compared to a cooperative or isolated, i. In addition to investigating go—nogo Simon task performance in an independent, cooperative and competitive joint condition, the present study made use of an individual go—nogo task at the beginning of the experiment to provide a valuable reference for the resulting JSEs.
Furthermore, the definitions of the terms cooperation and competition were further adjusted from those used by Iani et al. Here, cooperation instructions emphasize team work for achieving a common goal and not a cooperative competition against unspecified others as in Iani et al.
To further amplify the effect of interactive contexts, reward was given for every correct and fast enough trial. More importantly, however, the present study followed the recommendation of Liepelt et al. Liepelt et al. That is, compatibility effects like Flanker, Simon, and Stroop are typically smaller after incompatible compared to compatible trials Gratton et al. This conflict adaptation or Gratton effect is considered to reflect reduced interference as a consequence of cognitive control already being up-regulated in the trial following an incompatible conflicting trial Botvinick et al.
These can either be i nogo—go transitions, where the participant had to withhold a response in the previous trial but is required to respond in the current trial, or ii a go—go transition, in which the participant was required to respond in both the previous and the current trial.
Interestingly, while sequential adaptation effects were stronger for nogo—go transitions than for go—go transitions in both tasks, these where overall smaller in the individual go—nogo task suggesting additional between-person discrimination i. Considering a positive to neutral and thus, rather cooperative style when engaged in social interactions with others as default Iani et al.
In other words, do participants in a competitive relationship apply a self- or other-referenced focus Poortvliet and Darnon, The only thing they can influence in such a situation is their own action, which should result in a self-referenced focus Iani et al.
Based on this framework, the present study investigated the processes underlying flexible adjustments to the contextual challenge of either cooperating or competing for reward when interacting with others. Prior to the final part of the experiment participants were informed that the amount of reward will be equally divided between both participants i.
If participants in the competition group develop an other-referenced focus, we would expect larger JSEs and sequential processing adjustments after nogo—go transitions compared to the cooperative group.
On the other side, if participants in the competitive group apply a self-referenced focus, they should show a smaller JSE, and larger adjustments after go—go transitions compared to the cooperative group. Results in the individual go—nogo Simon task and the independent joint go—nogo Simon task should be in line with previous findings showing no go—nogo Simon effect but significant sequential processing adjustment in the former and significant effects in both measures in the latter condition.
Participants gave their written informed consent before their inclusion in the study in accordance with the ethical standards of the German Psychological Society DGPs; and the Declaration of Helsinki. No such requirements were present for this study. After the session, all participants were debriefed and rewarded with partial course credit.
Participants were tested in pairs and did not know each other prior to the experiment. Data from three participants were excluded due to mean reaction times or error rates of more than 2. For the present go—nogo Simon tasks, a green and a blue circle with a diameter of one centimeter were used as stimuli 0. Hommel et al. They were presented 8.
Upon arrival at the laboratory, pairs of participants were informed about the three consecutive segments of the experiment, namely performing the first task alone and the following two together with the other person. That is, while the participant assigned to the left workspace was seated in the left chair and responded via the left response key i.
Both participants were instructed to put their right index finger on the respective response key while leaving their left hand underneath the table on their left thigh. Experimental setup in the individual go—nogo Simon task Task 1; A and in the joint go—nogo Simon tasks Tasks 2 and 3; B. In both go—nogo Simon task contexts A,B the participants are required to respond to their assigned stimulus blue circle, person on the right, incompatible trial; green circle, person on the left, compatible trial by operating the response key in front of them.
Stimulus—Response assignments as well as spatial position of the participants were counterbalanced across participants but held constant across the tasks. Whereas in the individual go—nogo Simon task 1; A participants worked on adjacent computers, in the joint go—nogo Simon tasks 2 and 3; B both participants sat in front of one computer.
After the instruction phase was completed, the experimental phase of Task 1 started. There were two blocks of trials, which equally often contained each stimulus blue vs. This task was used to calculate the individual reaction time RT threshold for performance-contingent reward receipt in Task 2 and Task 3.
The threshold was determined by the 0. To maintain vigilance throughout the whole experiment, short self-paced breaks between blocks and a 2-min break between Task 1 and Task 2 outside the laboratory were provided.
Following Task 1 and a recovering break, participants reentered the lab to continue with the second segment, a joint go—nogo Simon task with performance-contingent reward. In order to keep S—R assignments and responsibilities consistent with Task 1, both participants were asked to take their respective seat of either the left or the right workspace counterbalanced across pairs of participants.
Thus, while the workspace remained the same for one participant, the other had to change, but the spatial assignment of chair and response-key remained the same see Figure 1B. After participants were reminded about stimuli and respective assignments, they were instructed about the possibility of earning four cents for every correct and very fast response i.
Note, however, to keep the task fair, a participant would lose two cents in case of an error and the partner would gain these two cents, because an error of one participant always represented a lost opportunity for the other participant to gain reward.
After the instruction, participants performed 16 more training trials in order to get familiar with the task and to give the participants a feeling of about how fast they have to react to receive the reward. Following this short training, participants got feedback about the amount of money they would have received before the experimental phase of Task 2 started.
As in Task 1, participants had to perform testing trials divided in two blocks and they received feedback about the earned amount of money after each block. After Task 2, participants continued with the third segment, again a joint go—nogo Simon task with performance-contingent reward. You can view details and manage settings at any time on our cookies policy page.
Manage settings. Research projects The impact of competition and cooperation on the performance of a retail agglomeration and its stores.
The impact of competition and cooperation on the performance of a retail agglomeration and its stores. Overview The highlights of the paper are: Competition and co-operation between stores increase agglomeration performance.
0コメント