Date created: | Last Updated:

Category: Project

Description: The Chains of Habit: Repeated Coordination in Joint Decision-Making Elicits a Sense of Commitment Abstract: We will test the hypothesis that repeated coordination with a partner can elicit a sense of commitment, leading people to resist tempting alternatives and thereby sustaining cooperation through fluctuations in individuals’ interests. In our paradigm, participants will perform a repeated joint decision-making task either with the same partner (Partner Condition) for all trials, or with a different partner on each trial (Stranger Condition). When both players coordinate on the same option, both are rewarded. On some trials, participants are offered outside options presenting varying degrees of temptation to defect. Participants will be informed that they are coordinating with a partner or partners who are in the lab with them. We predict that the results of their choices, as well as the trajectories of their mouse movements, will indicate that participants in the Partner Condition will be more resistant to the temptation to defect. Keywords: coordination, decision-making, commitment, cooperation Introduction The prevalence and flexibility of human cooperation is unparalleled by any other species. We routinely work together to achieve ends that we could not achieve alone, even setting aside short-term interests to maximize the benefits to our interaction partners and larger social groups. In recent decades, a great deal of research in evolutionary theory, experimental economics and psychology has been devoted to investigating the evolutionary origins of human cooperation (Henrich & Henrich, 2007; Nowak, 2012; Tomasello, 2009; West, Griffin, & Gardner, 2007). This has led to significant progress in specifying ultimate (i.e. evolutionary) mechanisms that are likely to have supported the evolution of cooperation in humans – e.g. kin selection (Hamilton, 1963; Maynard Smith, 1964), direct (Trivers, 1971) and indirect (Nowak & Sigmund, 1998) reciprocity, and cultural group selection (Boyd & Richerson, 2008). Moreover, this research has also informed and constrained research into the cognitive and motivational mechanisms that proximally support cooperation. For example, theorizing at the evolutionary level about the importance of indirect reciprocity has inspired research at the psychological level devoted to illuminating the mechanisms by which people manage their reputations (Nowak & Sigmund, 1998; Fehr et al., 2002). Thus, it has shown that people behave more pro-socially when they believe they are being observed, or when they believe that they will interact again with the same partner in the future (Andreoni & Bernheim, 2009; Rege & Telle, 2004). Moreover, it has been hypothesized that reputation management may be subserved by prosocial preferences, such as a preference for fairness (Andreoni, 1990) or an aversion to inequity (Fehr & Schmidt, 1999). Alternatively, Dana et al. (2007) and Heintz et al. (2015) have proposed that an aversion to disappointing others’ expectations may play an important role. Recently, researchers (Rusch & Luetge, 2016; Tomasello et al., 2012) have begun to investigate the psychological implications of a different theory about the evolutionary origins of human cooperation, namely Roberts’ (2005) ‘interdependence hypothesis’. According to this theory, humans’ tendency to cooperate arose evolutionarily in a period in which our ancestors lived in small groups of individuals whose interests were largely interdependent, and for whom it was therefore not typically beneficial to act selfishly to the determinant of other group members. In other words, human cooperation is a byproduct of earlier selection for skills in coordinating. From this starting point, Rusch & Luetge (2016) reason that humans may be equipped with social decision-making processes which do ‘not differentiate between instances of coordination and cooperation too sharply, at least initially, reflecting the hypothesized predominance and earlier evolutionary solution of coordination problems in our ancestral social ecology’ (Rusch & Luetge, 2016: 292). If so, then successful coordination with a partner in a repeated coordination game may lead people to view their partner as being a reliable partner in general, and therefore also as someone who is likely to resist temptations to behave selfishly (e.g. to cooperate in a prisoners’ dilemma). As a result, they themselves should be more likely to cooperate with a partner with whom they share a history of successful cooperation. In support of this, Rusch & Luetge (2016) found evidence of a ‘spillover effect’ from coordination to cooperation, i.e. cooperation rates in a prisoners’ dilemma were boosted when rounds of the prisoners’ dilemma were interspersed among rounds of a coordination game (i.e. the stag hunt) played together with a fixed partner. But while Rusch & Luetge’s hypothesis explains why successful coordination may lead participants to trust that their partner will not defect, it does not directly explain why they themselves would then choose to cooperate. Thus, it can only explain why cooperation rates would not fall below the level that corresponds with participants’ preference; it does not provide any explanation of why coordination may directly boost cooperation rates. And indeed, since cooperation in a standard prisoners’ dilemma depends not only on participants' trust that their partner will cooperate but also on their own willingness to cooperate, it is difficult to determine whether successful coordination in this paradigm boosted cooperation rates by affecting the former, the latter, or both. One way to isolate the sense of commitment to one’s partner – i.e. to specifically assess the effects of coordination upon the willingness to cooperate independently of trust – is to implement a one-sided social dilemma, such as a dictator game, in which only one player is faced with a temptation to defect. Since the decision of the dictator in a dictator game fully determines the outcome, she does not need to trust in the good will of her partner. This is the strategy employed by Guala & Mittone (2010). Their aim was to test the hypothesis that conventions take on a normative character over time. To this end, they measured cooperation rates in a sequential, one-sided prisoner’s dilemma (effectively a dictator game) that followed several rounds of successful coordination in a pure coordination game (choosing one of two colors, ‘red’ or ‘blue’), and compared this with cooperation rates in the same sequential, one-sided prisoner’s dilemma played without any prior coordination game. Their results confirmed this prediction, corroborating the hypothesis that repeated coordination can give rise to social norms which, in turn, function to stabilize cooperation. However, the results of this study are also consistent with an alternative explanation, namely that the coordination phase may have led participants to resist the temptation to defect by eliciting prosocial attitudes. This conjecture is motivated by research showing that coordination can enhance rapport (Bernieri, 1988) and trust (Launay et al., 2013; Mitkidis et al., 2015), and lead to cooperation in social dilemmas (Wiltermuth & Heath, 2009; Van Baaren et al., 2004) as well as pro-social helping behavior (Kokal et al., 2011; Valdesolo & Steno, 2011). If this conjecture is correct, then players may also have been more likely to cooperate if the choices presented in the test trials (the prisoner’s dilemma) differed from the choices offered in the coordination phase, i.e. if the cooperative choice did not correspond to the convention that had been established. Moreover, since participants in both conditions played in fixed groups, it is not clear whether the coordination phase boosted their willingness to cooperate in general, or by eliciting a sense of commitment only to their specific partners. The Current Research Building upon the findings reported by Guala & Mittone (2010) and Rusch & Luetge (2016), the current study investigates the effects of repeated coordination with the same partner upon people's willingness to cooperate with that partner. Like Rusch & Luetge (2016), we reason that if Roberts' (2005) interdependence hypothesis is correct, then repeated coordination with the same partner should boost cooperation. Unlike Rusch & Luetge (2016), we reason that the interdependence hypothesis generates this prediction independently of any effects of coordination upon trust. This is because repeated coordination may serve as a cue to participants that their partner is interdependent with them, engendering a sense of commitment to that partner and thereby making them more resistant to tempting outside options (Michael, Sebanz & Knoblich, 2016). In order to isolate the effects of repeated coordination on the willingness to cooperate with one's partner irrespective of trust, we will follow Guala & Mittone’s (2010) approach in opting for a sequential, one-sided social dilemma on test trials. We have also modified their design in two key respects. First, unlike their paradigm, ours compared a condition in which participants played in fixed pairs with one in which they had a different partner on each trial. Specifically, in one condition, participants play a sequential coordination game with the same partner (Partner Condition) for 126 rounds, while in a second condition the participant plays the same game with a different partner on each trial (Stranger Condition). This will enable us to tease apart any general prosocial effect of coordination from the effect arising from a sense of commitment specifically to one’s partner. Secondly, we aim to ensure that any effects of coordination upon cooperation rates could not be explained as adherence to specific conventions that had arisen during the experiment. To do this, we designed the coordination game such that the values and colors varies from one trial to the next, such that the cooperative option does not reflect any convention. Participants will be informed that they are coordinating with their partner(s) via internet, and that their partner(s) will not receive any feedback about any of their decisions until the end of the experiment. At the beginning of each round, an image of the partner’s player number will be displayed, which is either the same (Partner Condition) or different (No Repetition) every round. Then, the partner chooses one of two values. The participant does not see what these two values were, but was then herself presented with two values to choose between. One of these, indicated in green or blue, is the same value that the partner had chosen (cooperative option); the other, indicated in orange, is an alternative value (alternative option). The value of the alternative option varies unpredictably across trials; the degree of temptation which it presented was a function of its value. We predict that participants would resist the tempting alternative option more frequently in the Partner Condition than in the Stranger Condition. We also reason that a greater sense of commitment in the Partner Condition will lead participants to experience a higher degree of cognitive conflict on those trials on which they do choose the alternative option, and a lower degree of cognitive conflict on those trials on which they cooperate. To measure cognitive conflict, we will analyze the response dynamics using continuous tracking of mouse movements while participants chose between the two options, which were spatially separated on the screen (Kieslich et al, 2014; Freeman, Dale and Farmer, 2010; Spivey & Dale, 2006; Spivey, Grosjean & Knoblich, 2005). Method Participants Sample Size: Small-Medium sized effect (Cohen's d=.4 n= 52). This enables us to collect the data in two sessions of 24. We will book in 60 people, assuming some no shows, and then include all of the data that we collect in the analyses. Apparatus and Stimuli The experiment is displayed on a 24-inch wide screen (16:9) computer monitor (resolution: 1920 x 1080 pixels, framerate = 60Hz.). The program for the experiment is written in Open Sesame (Mathôt, Schreij, & Theeuwes, 2012). The two choice options are presented as 7 cm x 5 cm rectangular fields, separated by 36 cm. The mouse start field is also 7 cm x 5 cm, and is positioned 18 cm lower at the midpoint between the two choice options. Procedure Trial Structure. First, a picture of the partner’s number is displayed for 2000 milliseconds (a). Next (b), the participant waits as the partner selected one of two values, which are not displayed to the participant (b) The duration was determined randomly and ranged from 2000 to 3000 ms. Then, the two colored boxes appear at the top of the screen. Only when the participant moves the cursor from the mouse start field did the value appear in the colored boxes (d). Each trial ended when the participant clicked on one of the two options or when 2999 milliseconds had elapsed. After the participant has made her or his selection, there is a 500 ms delay, and then the payoffs are displayed for 3000ms. There are 80 test trials in each block (plus 10 practice trials). Each block corresponds to one condition (Partner / Stranger). Predictions: - Higher cooperation rate in partner condition - More foregone earning in partner condition - Higher RTs when defecting in the partner condition - more pull toward cooperate when defecting in the partner condition - more AOC when defecting in the partner condition - Higher RTs when cooparating in the stranger condition - more pull toward defect when cooperating in the partner condition - more AOC when cooperating in the stranger condition