This item is only available as the following downloads:
THE ROLE OF COMPUTER-BASED CONTEXT IN COOPERATIVE, COMPETITIVE, AND INDIVIDUALISTIC DECISION-MAKING BY RACHEL SCHERER A Thesis Submitted to the Division of Social Sciences New College of Florida in partial fulfillment of the requirements for the degree Bachelor of Arts Under the sponsorship of Dr. Heidi Harley Sarasota, Florida May, 2011
Acknowledgements I would like to thank my thesis sponsor Dr. Heidi Harley for her direction, critical eye, and support throughout the thesis process. I would also like to thank Dr. Steven Graham and Dr. Charlene Callahan for being on my committee, and fo r acting as resources as my thesis evolved. I would like to thank Professor Cooper for all of the time he spent assisting me with the analysis of my data, and for maintaining my favorit e thesis hideaway, the SSRL. I would also like to extend my deep gratitude to David Turon, wh o lent me his critical perspective and was invaluable in the collection of my data. Finally, I would like to thank my peers in Psych S enior Seminar, my family, my roommates, and the staff at the University Parkway Starbucks, for all of their endless support and encouragement.
ii THE ROLE OF COMPUTER-BASED CONTEXT IN COOPERATIVE, COMPETITIVE, AND INDIVIDUALISTIC DECISION-MAKING Rachel Scherer New College of Florida, 2011 ABSTRACT The tendency to choose either to cooperate or compe te can be affected by many factors, including ones individual predisposition, the situ ational significance of the decision, and the context in which one makes the decision. Increasing ly, collaboration and interaction via the internet are becoming more commonplace. The current study seeks to determine whether the context of a computer-based interactionexplicitly cooperative or explicitly competitivecan affect an individuals choice to cooperate or compe te when that individual is no longer immersed in that context, or, in other words, has re-entered real life, similarly to how context can affect decision making in offline scenarios. In this study, nearly all participants reflected an individualistic social value orientation after participating in the computer-based context m anipulation, regardless of their measured social value orientation in the predisposition task prior to the context manipulation. Participants decision-making did not end up being in alignment w ith the social value orientation condition to
iii which they were assigned; the computer-based nature of the task itself served as sufficient context to manipulate participants decision-making toward the individualistic. __________________________ Dr. Heidi Harley Division of Social Sciences
iv TABLE OF CONTENTS Acknowledgements Abstract . Table of Contents .............................................. ......... List of Figures and Tables . Introduction . Cooperation and Competition in Decision-Making Context and Social Value Orientation . Assessing Social Value Orientation .. Social Value Orientation Self-Report Objective Performance Measures Social Dilemmas . Decomposed Choice Tasks .. Priming Social Value Orientation . Decision-making in a Virtual Arena The Present Study Method Participants .. Materials .. Tasks Comparing the Predisposition Task and the RMSVO Stimuli .. Procedure i ii iii iv 1 2 2 4 4 5 5 6 9 12 14 15 15 16 16 19 20 23
v Results .. Discussion Computer-Based Context and Social Value Orientation . Face to Face versus Computer Mediated Communication .. Implications in Web-Based Crowdsourcing............ ................................ Implications in Web-Based Education... ............ ................................. Conclusion .. Appendix A Appendix B . Appendix C . Appendix D Appendix E . References 25 28 28 30 31 32 33 34 40 41 45 49 50
vi List of Figures and Tables Figure 1: Graphical Output from the Ring Measure of Social Value Orientation Figure 2: RB-830 Response Pad Figure 3: Example Stimulus Item from the PANAS-X Figure 4: Example Stimulus Item from the Predisposi tion and Ring Measure Tasks Figure 5: Example Stimulus Item from the Context Ma nipulation Task Figure 6: SVO Angles from X-Axis, By Condition Figure 7: SVO Preand Post-Test Mean Angles from t he X-Axis, by Condition
Social Decision Making in a Computer-Based Context | 1 The decision either to cooperate or compete can be affected by many factors, including ones predisposition toward a given socia l value orientation alignment (Hertel & Fielder, 1994; Hertel & Fielder, 1998), ones per ception of how one is perceived by others (Komorita et al., 1980; Loch et al., 2006), the situational significance of the decision(Drouvelis et al, 2010; Eiser & Bahvnani, 1 974, Tremayne & Curtis, 2006; Zineldin, 2004), and verbal priming tasks (Hertel & Fielder, 1994; Hertel & Fielder, 1998). Increasingly, collaboration and interaction via the internet are becoming more commonplace, and this change in arena has also brou ght to light differences in how people interact in a computer-based setting versus in real life. For example, people engage in massively multiplayer online games, such as World of Warcraft or Eve Online, specifically in order to seek out opportunities to interact with others in new or unexpected ways and to have opportunities to coop erate with others which are not present in real (versus virtual) life (Frostling-He nningsson, 2009). Because computers have become a venue both for cooperation and compet ition through the advent of such phenomena as online collaboration, gaming, social n etworks, and group editing of wikis, it is of interest to determine whether the contextu al quality of a computer-based interactioni.e., explicitly cooperative or explic itly competitivecan affect an individuals choice to cooperate or compete when th at individual is no longer immersed in that context. Thus, my question: Does context i n a computer-based social interaction priming task affect the level of cooperation or com petition exhibited outside the task? The answer to this question could help us to unders tand better how our choice behavior can be manipulated by the contexts in which we imme rse ourselves.
Social Decision Making in a Computer-Based Context | 2 Cooperation and Competition in Decision-Making In social situations, it is common to find oneself choosing whether to engage in cooperative (pro-social) behavior, competitive beha vior, or individualistic behavior. These decisions, while not mutually exclusive (Loch et al., 2006; Zineldin, 2004), are frequently at odds with each other. Typically, when choosing to engage in cooperative behavior, one maintains the goal of optimizing the outcome afforded to a group over ones individual outcome. The reverse is true for c ompetition: in choosing to compete, one characteristically is explicitly attempting to optimize ones own outcome over the potential outcome of a group or another individual (Komorita et al., 1980). Individualistic attitudes, by contrast, are marked by a goal of opt imizing ones own outcome without reference to the actions or outcomes of others in a group. There are many motivations for one to choose to cooperate with or compete against others in a social situation. Cooperative behavior may be inherently rewarding: t hat is, people are sometimes motivated to choose particular courses of action fo r the sake of cooperating (FrostlingHenningson, 2009). The decision to behave cooperati vely or competitively when interacting with others can be influenced by variou s environmental factors (FrostlingHenningson, 2009; Hertel& Fielder, 1998; Tremayne & Curtis, 2006) including ones mood, predisposition toward either competition or c ooperation, and the situational meaning of the decision (Eiser & Bahvnani, 1974; He rtel & Fielder, 1994). Context and Social Value Orientation Eiser and Bahvani (1974) found that situational mea ning or context can affect ones decision either to cooperate or compete in an iterated Prisoners Dilemma task,
Social Decision Making in a Computer-Based Context | 3 even when such factors as communication opportuniti es, reward magnitude, and partners strategy were held constant across conditions. The iterated Prisoners Dilemma task is a form of social dilemma which places players in a sc enario of being an apprehended criminal, and requires that they choose whether to make a plea, landing their partner in jail but reducing their own jail sentence, or to st ay silent about the crime, which might result in both participants serving no jail time if both choose to behave pro-socially, or might result in a longer jail sentence for the play er who stays silent if the other defects. Participants were seated at a desk separated visual ly from the other participants and told that they had been paired with another individual i n the room to play the Prisoners Dilemma game, but that the identity of that individ ual would not be revealed. Context was manipulated through instructions which differed across four conditions, and which indicated the way in which the results of the task would be interpreted. The conditions were described as Control (Co), Economic Bargaining (EB), Interpersonal Interactions (II), and International Negotiation (IN). It was pr esumed by the experimenters to be implied that in the II and IN conditions, it would be more favorable to cooperate (due to assumptions that most people would find it importan t to act in a friendly manner and to keep the peace internationally, respectively) tha n it would be in either the Co or EB conditions (in which participants might be more ori ented toward personal gain either due to the economic bargaining context or due to the ve ry nature of the Prisoners Dilemma). This context manipulation was problematic because i t assumed certain values to be held in association with different types of interaction (i.e., economic, political, or interpersonal), which could possibly have been subj ective, although the results were in alignment with the researchers expectations. Situa tional meaning did affect participants
Social Decision Making in a Computer-Based Context | 4 levels of cooperation: the mean percentage of coop erative responses in the II and IN conditions were 52% and 58% respectively, as compar ed to 42% and 43% in the Co and EB conditions respectively, suggesting that those c ontexts in which cooperation was more highly valued encouraged more cooperation in p articipants. Assessing Social Value Orientation The tendency to make cooperative decisions might be indicative of an individual having a generally pro-social social value orientat ion. Social value orientation motivates how people make decisions in game scenarios. Messic k and McClintock (1968) identified different social value orientations, such as pro-so cial, competitive, and altruistic. Social value orientation can be measured through a variety of means, and can be described either in terms of an individuals state or traits. That is, a longitudinal view of decisions which are fairly consistent with a particular socia l value orientation can reflect a particular long-term attitude, or predisposition, w hile a short-term analysis can reflect an individuals current, state orientation. Different methods of assessing social value orientation are outlined below. Social Value Orientation self report. Self-report measures are frequently used to assess social value orientation. These measures require participants to express the degree to which they would cooperate or compete with others, either before, during, or afte r imagining some interaction or scenario (Arnocky et al., 2007; Engelmann et al., 2 010; Loch et al., 2006). Self-report measures are useful in that they give participants the opportunity both to make a decision and, frequently, to qualify it, however participant s reports might not be consistent with
Social Decision Making in a Computer-Based Context | 5 the participants behaviors. Behaviors can provide more objective measures of performance. Objective performance measures. Objective performance measures in which behaviors a re coded as being in alignment with particular social value orientations can be highly reflective of reality in that they involve actual observation of behavior, b ut they sometimes fail to capture the intent of the participant. For example, such questi ons might arise as, Did this individual mean to engage in a behavior which was anti-social? What was his motivation? What affected his actions?(Loch et al., 2006).Examples of objective performance measures include social dilemmas and decomposed choice tasks Social dilemmas. Dilemma problems can assess Social Value Orientatio n (Liebrand, 1988). Dilemma problems pose a scenario in which a player must make a decision that will affect both her and another player or multiple play ers outcomes. These measures have the advantage of allowing researchers to look at th e actual choices of the individual rather than relying on self report. Due to their narrative nature, social dilemma problems tend to skew the context of the posed interaction in terms of whether competing or cooperating might be more socially desirable. Thus, these measu res can be problematic to use on their own to determine social value orientation. Examples of social dilemmas include the Prisoners Dilemma and the Stag Hunt Dilemma, a dil emma task similar to the Prisoners Dilemma described earlier. In this task, participan ts choose between safety and social cooperation through deciding whether to hunt a stag which necessitates the cooperation
Social Decision Making in a Computer-Based Context | 6 of another player in order to succeed, or a hare, w hich a player can successfully hunt alone but which yields a substantially lower payoff Decomposed choice tasks. Another kind of measure, the Decomposed Game or Dec omposed Choice Task, utilizes a series of related dilemmas which measure a parameter known as motivational orientation. This parameter represents the relative weights in terms of both attitude (generally altruistic/individualistic) and choice b ehavior (generally cooperative/competitive) that a participant assigns to his own and to anothers outcomes (Griesinger & Livingston, 1973; Pruitt, 1967). Thes e weights can be used to determine an individuals orientation toward cooperation or comp etition. Some measures, for example the Ring Measure of Social Values, use an approach which combines a series of dilemmas with a measurement of response latency. Me asuring response latencies between the time that a choice is presented and a d ecision is made can also apply a value to represent the relative strength of each decision : for example, when the decision to compete is consistently made a number of seconds mo re quickly than one to cooperate, it can be assumed that a participant is more inclined to compete or finds it easier to make the decision to compete at the time of assessment ( Hertel & Fielder, 1994; Liebrand & McClintock, 1988, Murphy et al., 2010). Decomposed Choice tasks, such as the Ring Measure o f Social Values, combine the advantages of using dilemma scenarios to assess social orientation with the cognitive measure of evaluating response latencies to assess strength of orientation. The Ring Measure consists of twenty-four pair-wise decisions which require the participant to
Social Decision Making in a Computer-Based Context | 7 allocate some amount of money (whether it is some a mount of gain or loss) to herself and to one anonymous other. For example, a participant might be asked to choose whether to allocate a loss of eight dollars to herself and a l oss of twelve dollars to the other, or to allocate a loss of eight dollars to herself and a l oss of four dollars to the other. Each decision yields a pair of coordinates which describ e (in [x,y] format) ones orientation toward the self (measured along the x axis) and one s orientation toward some other (measured along the y axis). These coordinates allo w the data to be represented graphically in aggregate (see Figure 1 for a graphi cal representation of Ring Measure output) to show a participants social orientation toward cooperative or competitive behavior. This visual representation is valuable be cause it works as a vector, describing not only valence but also magnitude: For example, i f the line described by the coordinates is long, it can give a measure of stren gth to the result because many decisions which benefit the self would yield many positive va lues in the x-coordinate and thus a longer vector. On the other hand, inconsistent choi ce behavior would yield both positive and negative values in the x-coordinate, visibly re ducing the length of the vector and giving a graphical measure of strength.
Social Decision Making in a Computer-Based Context | 8 Figure 1 Graphical output from the Ring Measure of Social Va lue Orientation
Social Decision Making in a Computer-Based Context | 9 Priming Social Value Orientation Much research has been done on priming various soci al value orientations, for example, competitive and cooperative. Priming a SVO entails tweaking an individuals tendency toward a certain type of deci sion-making, through indicating that a certain type of behavior might be more socially des irable given a particular situation or through verbal priming. Hertel and Fielder (1998) f ound that verbal priming, a type of priming which relies on positivelyor negatively-v alenced word associations to induce positive or negative associations with a concept, c an be used to affect ones choice to compete or cooperate, and that an individuals posi tive or negative mood can correctly account for the amount of variability in choice beh avior that is exhibited by the participant (i.e., whether the participant sticks w ith either cooperative or competitive choices fairly consistently, or whether instead the participant chooses to switch the types of choices he is making from decision to decision). In this study, participants were subjected to a mood manipulation in the form of one of two short films which avoided the themes of cooperation or competition and were inten ded to induce either a positive or negative affect in the participants. Then, the part icipants engaged in a verbal learning task wherein they were instructed to memorize as many of thirty words scattered on a page as they could, in order to report them later. Each set of words contained three subsets of ten spatial words (such as open or skewed), ten f iller adjectives (such as blue or healthy) and ten words which were intended to rep resent either negative connotations of cooperation and positive connotations of competi tion, or vice versa (e.g., inhibited and weak would contrast with self-confident and triumphant in a condition which primes for competitive behavior). Finally, particip ants completed a four-person dilemma
Social Decision Making in a Computer-Based Context | 10 game scenario which posed a problem about sharing a fictional aircraft among the members. The aircraft would require maintenance fro m the group members in order to remain functional, and in order to choose to cooper ate with the other players participants had to choose to give up resources that they were a llocated. Participants were also administered mood manipulation checks as well as as ked to report the words with which they had been presented at the beginning of the exp eriment, as well as any themes they noticed with those words. When, using a verbal prim ing task, either positivelyor negatively-valenced connotations were made in conne ction with cooperation or competition, that attitude which was given positive connotations was more likely to be chosen by participants. These results suggest that participants are able to differentiate between SVO-aligned decision types in their decisio n-making, and that they will respond to the type of context in which a concept is introd uced: for example, when competition is activated as a positively-connoted course of action s through verbal priming, it becomes more likely that the participant will behave compet itively. Drouvilis et al. (2010) tested the hypothesis that activating the concept of cooperation through social priming techniques leads to pro-social behavior, and found that priming cooperation leads to increased group c ontribution levels from individuals. This phenomenon occurred most notably in females, l ikely due to differences in how men and women perceive group contributions. Participant s were either entered into the NP, or neutral prime condition, or the P, or positively prime condition. Participants were primed through the completion of a word search at t he beginning of the experimental session. A list of 20 words was embedded in a 16x16 matrix. Both treatments contained the same 5 neutral words (carpet, lamp, plant, sham poo, window). In the P-treatment, the
Social Decision Making in a Computer-Based Context | 11 remaining 15 words were related to cooperation. In the NP-treatment, the remaining 15 words were not related to the concept of cooperatio n. Participants were told that their performance on this task would not impact their ear nings in the subsequent task. Then, participants completed a one-shot linear public goo ds game. The linear public goods game was identical in both conditions. The game inv olved participants being placed in groups of three and given 20 tokens each, of which they decided how many to contribute to the group pot versus keep to themselves. Keeping tokens to oneself resulted in a return of 1 money unit per token; adding tokens to the gro up pot resulted in a return of 0.5 money units per group member, or 1.5 units total gr oup benefit. Thus, there is an economic incentive for the more self-oriented parti cipant to retain as many tokens as possible, since individual return for a group contr ibution comes at a net cost of 0.5. Alternatively, the more socially-oriented participa nt might consider that if all participants contribute all of their tokens to the group pot, th e individual income ends up being 30 tokens, which is much greater than her capital. Aft er completing this task, participants reported on their beliefs about the contributions o f other group members, and on the magnitude of positive or negative affective respons e to those contributions. Contributions of either 0 or 10 tokens were modal in both treatme nts; however, only in the P-treatment did some participants contribute all 20 tokens from their initial endowment. There was no significant difference in level of contribution bet ween the two conditions, but this was explained through gender differences in contributio n, i.e., women contributed to the group significantly more than men in all conditions and those women who were positively primed believed that their group members would contribute more than did the men believe. It followed that women who were positi vely primed ended up contributing
Social Decision Making in a Computer-Based Context | 12 to the group more than those women who were not, su ggesting that these beliefs about others contributions (which were elicited through priming) guided decision-making in making individual contributions. Decision-Making in a Virtual Arena Computer-based and online venues are becoming utili zed more frequently as a medium for social interaction, and, increasingly, c ollaboration and interaction via the internet is becoming more commonplace. This change in arena has brought to light differences in how people interact in a computer-ba sed setting versus in real life. People interact in many varied ways using the computer as a medium, e.g., collaboration on documents or projects; online gaming; contributions to wikis, forums, or other web-based information hubs; and interaction and gaming throug h web-based social networks such as Facebook or MySpace. People are motivated to use some of these computer -based or online facilities because they are venues for social interaction (Fro stling-Henningsson, 2009). For instance, on average, people who play massive multi player online games (also known as MMORPGs) tend to rate interaction with others as their highest motivation for engaging in these types of games (Frostling-Henning sson, 2009; Wu et al., 2010). Those with a strong cognitive preference for a virtual (v ersus real) life tend to gravitate toward online multiplayer games, and the preference for a virtual life has a negative correlation with real-life social skills (Liu &Peng 2009). Frostling-Henningsson (2009) found that among moti vations to participate in MMORPGs, the opportunity for cooperation and com munication rated highest
Social Decision Making in a Computer-Based Context | 13 among those surveyed. Interviews were conducted wit h young adults from game centers in Stockholm to determine the reasons they had for participating in these games. Players of two games in particular, Counterstrike and World of Warcraft, were targeted for participation in this study. Participants cited bei ng able to connect with others in unexpected ways or to cross boundaries (i.e., physi cally having the ability to do things in the game scenario that would not be possible in rea l life, such as casting magical spells or flying) as motivational reasons to interact with ot hers in MMORPGs. These findings lead to another question, which would be whether those p articipants who tended to value opportunities to cooperate highly as characteristic of their participation in MMORPGs also exhibited a predisposition toward cooperative behavior in real life (a.k.a. outside of the game scenario). This question is one of many le ft unanswered by the gap in the literature about human decision making and computer -based interaction. As noted earlier,Eiser and Bahvnani (1974) began to address the question of computer-based interaction and choice, but the priming method used was simply a manipulation of the instructions offered rather than an immersion in an explicitly cooperative or competitive computer-based environment, e.g., a MMORPG such as Counterstrike or World of Warcraft, or an online document collaboration effor t. Derks, Bos, andGrumbkow (2004) found that participa nts use more emoticons in online text-based chat interactions in socio-emotio nal versus task-based contexts, and, further, that participants used more emoticons when the valence of those situations was positive, rather than negative. Participants were p resented with a questionnaire which showed them a series of short internet chats, which were classified as having one of two social contexts: either task-oriented, in which the subject of the conversation was the
Social Decision Making in a Computer-Based Context | 14 division of tasks in an important school project, o r socio-emotional, in which the subject of conversation was to brainstorm with a friend abo ut what sort of present to buy a mutual friend. Each context was also given either a positive or negative valence. Participants were asked to respond to every chat ei ther with text, an emoticon, or a combination of both. Frequency of emoticon use was measured, and researchers interpreted the results as showing that emoticons a re more likely to be used in socioemotional contexts, thus as a means for conveying e motion and making up for the lack of facial cues in CMC (Derks et al., 2004). This study also shows that participants distinguished between two types of computer-based c ontexts, treating, for example, a positively-valenced socio-emotional scenario much d ifferently than they treated a negatively-valenced task-oriented scenario. The cur rent study bases its method of contextual priming on the ability of participants t o make these computer-based contextual distinctions. The Present Study It is given that as humans, our social decision-mak ing is affected by a number of factors, including context (Drouvelis et al, 2010; Eiser & Bahvnani, 1974; Tremayne& Curtis, 2006; Zineldin, 2004), mood, predisposition (Hertel & Fielder, 1994; Hertel & Fielder, 1998), and verbal priming (Hertel & Fielde r, 1994; Hertel & Fielder, 1998). We are moving toward interactions in an entirely new v enue, the internet, and in this venue there are usually limited opportunities to take adv antage of many social cues such as facial expression. However, we maintain some of the same motivations to interact online or over the computer as we do in real life. Presuma bly, such factors as those listed at the beginning of this paragraph, which are internal to ourselves, are still at play when
Social Decision Making in a Computer-Based Context | 15 decision-making moves to the virtual world. The pre sent study investigates this assumption. The question posed in the current study was address ed through first assessing the participants state affect and predispositions towa rd or away from cooperation and competition prior to the manipulation. Participants then engaged in a context priming task which involved the participants being assigned to e ither an Individual, Cooperative, or Competitive condition, in which they were told that they would be completing a task for as many points as they could either alone, as p art of a team of four, or in competition against three opponents, respectively. In this task participants responded to 32 brainteaser questions. During the task, the time taken t o complete individual tasks was recorded. Then, participants completed the dependen t measure, a decomposed game called the Ring Measure of Social Value Orientation which assessed the participants state orientation toward one of eight possible SVOs :individualism, pro-socialism (cooperation), altruism, martyrism, masochism, maso -sadism, sadism, or competitionoriented, in decision-making. In this task, the res ponse latency was measured between the presentation of dilemma items and participant respo nses. Method Participants Participants were 48 students, aged 18-24, at a sm all liberal arts college in southwestern Florida. Eight participants were exclu ded from the study due to not having completed all required measures; therefore, data fr om 40 participants (27 women, 13 men) are included in this study. Participants were split into three groups: Individualistic (N=11), Cooperative (N=15), and Competitive (N=14).
Social Decision Making in a Computer-Based Context | 16 Materials Participants completed a series of computer-based tasks presented using SuperLab 4.0 (http://www.superlab.com/v4/) stimulus presenta tion software. Participant input was collected via a RB-830 USB response pad. (See Figur e 2 for an illustration of the response pad.) Figure 2 RB-830 Response Pad Tasks Participants first completed the Positive and Negat ive Affect Scale, then a social value orientation (SVO) task which served to gauge participants SVO predisposition, followed by a computer-based context manipulation a nd the Ring Measure of Social Value Orientation.
Social Decision Making in a Computer-Based Context | 17 The Positive And Negative Affect Scale (PANAS) assesses participants affect or mood through asking participants to assess their le vel of each of many attitudes on a fiveitem Likert scale (0 = not at all, 1= very sligh tly, 2=a little, 3=moderately, 4=quite a bit, 5=extremely). The PANAS-X was m odified to serve as a state measure through a slight change in the instructions : participants were asked to assess each item with respect to how they are feeling RIG HT NOW as opposed to being asked to think about the extent to which they had felt ea ch prompted emotion over the last two weeks. This measure served as a check on whether va riability in a given participants decision-making might have been affected by his moo d. If one is in a good mood, one is more likely to make decisions based on the strength of the choice (i.e., appropriateness to context) rather than to fall back on ones predispo sitions or make snap decisions (Eiser&Bahvnani, 1974). Thus, in an iterated functi on, decisions might look less consistent: one might agree to cooperate in one sit uation, and then immediately decide to defect thereafter. The Decomposed Choice Predisposition Task is a cooperative-competitive decomposed choice predisposition task used by Herte l and Fielder (1988). This task takes the form of a series of four pairwise two-player di lemmas. These dilemmas address state attitudes toward cooperation, competition, altruism and individualistic behavior through requiring a participant to make decisions about mon ey allocation to himself and to an anonymous other. Scoring and interpretation of thes e choices is similar to that associated with the Ring Measure of Social Values described below. The Context Manipulation takes the form of a computer-based game scenario i n which the participant plays a game with the goal of earning as many points as possible.
Social Decision Making in a Computer-Based Context | 18 The participant either plays alone (individual cond ition), or as part of a team with (Cooperative condition) or an opponent against (Com petitive condition) three other players. The game consists of a series of brain-tea sers including analogies, numerical pattern recognition, spatial pattern recognition, a nd odd-one-out tasks. In the Individual condition, the participant is tol d that each correct response yields him some points. Participants are not penalized for incorrect answers. In the Cooperative condition, participants are told that any correct a nswer on the part of the participant or his teammate earns the team a point. In the Competitive condition, participants are told that correct answers on the part of their opponent earn points for the opponent by taking points away from the participant. Participants in every condition were told that thei r goal was to earn as many points as possible. This context manipulation was designed to serve as a sort of attitudinal priming which aligned to cooperative, c ompetitive, or individualistic behavior, respective to the condition in which participants w ere involved. The Ring Measure of Social Values is a 24-item dilemma task which requires participants to make pairwise decisions about how a sum of money is allocated to oneself and to an anonymous other, and thus is presented as a two-player scenario. For each decision, participants are presented with two money -allocation options. Participants are not given any guidelines as to how money should be allocated (e.g., You should try to allocate as much money to yourself as possible), b ut rather are simply instructed to circle a choice with the knowledge that both the pa rticipants own and the anonymous others final score both depend on the choices that each of them make. These options
Social Decision Making in a Computer-Based Context | 19 vary in terms of whether the self or the other is p redisposed to win any given choice (that is, a particular decision might simply requir e the participant to choose how much more the other is allocated, rather than whether the other is allocated more). In this way, not only are cooperative and competitive actions as sessed, but so also are attitudes: when the situation is set up such that one player or the other gets the optimal outcome by default, is the participant more inclined to award the other a wider rather than smaller margin? The decisions made by the participant yield sets of coordinates which can be used to describe ones orientation toward decisionmaking influenced by oneself (x-axis value, determined by raw value of monetary allocati on to the self) versus influenced by an other or others (y-axis value, determined by raw value of monetary allocation to the other in the task). These coordinates can be summed up and plotted to create a vector whose directionality and strength can be interprete d to determine ones social orientation. In this administration of the Ring Measure, measure s of response latency werealso taken to add validity to data representing the strength o f decisions to cooperate, compete, or act in an individualistic manner. Comparing the Predisposition Task and the RMSVO The difference in social value orientation prior to and after the context manipulation was measured by asking participants to complete a social value orientation predisposition measure, which was derived from the Ring Measure of Social Value Orientation. This predisposition task contains diff erent and fewer questions than does the RMSVO, but the task was purported to be highly reli able with respect to the Ring Measure, and was used to measure change in social v alue orientation in the Hertel and Fielder (1998) study detailed above. To confirm the reliability of the results of the current
Social Decision Making in a Computer-Based Context | 20 study, the Ring Measure and the SVO Predisposition task were administered to a second group of participants. This group of 32 participant s completed a pen-and-paper version of the items from both the RMSVO and the SVO predispos ition task as a single task with the same set of instructions (see Appendix D for th e instructions used for both tasks). There was no indication that the items presented in this study were extracted from different measures, and, in the absence of any sort of manipulation between the two tasks, 87.5% of the participants in this task exhibited th e same social value orientation in both the Predisposition measure and the Ring Measure. Th ese data lend support to the validity of comparing these two measures in the current stud y. Stimuli Stimuli for each task were created using Adobe Illu strator CS5 (http://www.adobe.com/products/illustrator.html) so ftware. For the PANAS-X, these stimuli consisted of instructions and the presentat ion of individual words (see Figure 3 for an example stimulus item). For the Predispositi on and Ring Measure items, stimuli were created with a similar aesthetic to that used in the Computerized Ring Measure (Hertel& Fielder, 1988); i.e., choices are presente d as bar charts that compare the rewards to each player (see Figure 4 for an example Ring Me asure and Predisposition task item).
Social Decision Making in a Computer-Based Context | 21 ___________________________________________________ ___________________ Figure 3 Example stimulus item from the PANAS-X ___________________________________________________ ___________________ Figure 4 Example stimulus item from the predisposition and Ring Measure tasks ___________________________________________________ ___________________
Social Decision Making in a Computer-Based Context | 22 Figure 5 Example stimulus items from context manipulation ___________________________________________________ __________________ The items for the context manipulation were taken f rom online spatial IQ and pattern recognition resources (Brain Training Games 2011; Visual Patterns, 2011). (See Figure 5 for examples of task items.) Correct item answers were balanced across this section, with, for example, the answer number 0 o ccurring exactly eight times in this task: twice per each of four question types. Measur es of reaction time and performance accuracy were taken to confirm that participants we re engaged in the manipulation, but participant performance had no impact on the scorin g feedback that the participants
Social Decision Making in a Computer-Based Context | 23 received. This feedback was presented in the form o f animated performance bars which were presented at the end of every fourth task. Bec ause the feedback was not actually tied to participants performance, participants were dis couraged from attempting to keep track of their scores through a number of means. First, i n the social conditions (Cooperative and Competitive) participants were told that the ba rs rose and fell as a function of both the participants performance and that of the other players: i.e., in the Cooperative condition, the number of points displayed was said to be an aggregate total of the points earned by all four teammates in the game for good p erformance on the task items; similarly, in the Competitive condition, the number of points displayed on the meter rose and fell depending on the performance of the partic ipant and his opponents: the magnitude of increase or decrease was the net total of points earned by the participant when taking into account both his own performance ( doing well earned him points) and that of the other players (when the other players d id well, the participant was said to lose points). Secondly, participants were told that each type of question was worth a different number of points, but those amounts were not specif ied. Finally, to account for the issue of those in the i ndividual condition divining that the meters were not actually tied to their own perf ormance, all conditions were only presented with the point meter after every fourth r esponse. In this manner, it was difficult for participants to distinguish the number of point s which may have been awarded to them on any single, given task item. Procedure Participants were recruited via postings on a colle ge student forum, as well as on various social networking sites. Participants were told that their participation would take thirty minutes and would entail participating in an online game and making some choices
Social Decision Making in a Computer-Based Context | 24 about money allocation, as well as reporting on the ir current emotional state. For agreeing to take part in the study each participant received a $5 gift card for Starbucks. Participants were scheduled in time blocks, with th eir time slot matching at least one other participants where possible. Each partic ipant was assigned to one of three conditions: Individual, wherein the participant com pleted each task individually without being told that other players were involved in the game manipulation; Cooperative, wherein the participant was told that the context m anipulation task would be completed as a team effort, i.e., points would be rewarded in aggregate as a result of responses made by all four team members; or Competitive, wherein t he participant was told that in the game manipulation points would be awarded as a func tion of his own performance versus the performance of three opponents, i.e., that he w ould earn points for himself by taking them away from his opponents every time he got a qu estion correct, and that his opponents would also be able to take points from hi m by answering questions of their own correctly. A key part of this manipulation was that the oppone nts or team members did not actually exist. These team members or opponents wer e fabricated to increase the social salience of the context in which participants were immersed. Certain other features were added to make this deception more believable. For e xample, all participants were told that they could track the points rewarded to them b y looking at an animated point meter, which was presented after every fourth participant response. The animated point meter was actually not tied to participant response, sinc e it would then have to take into account actions from players that did not exist. Thus, the meter was made to (in the Cooperative and Individual conditions) rise arbitrarily every t ime it was presented. In the Competitive
Social Decision Making in a Computer-Based Context | 25 condition, the point meter rose and fell arbitraril y to simulate the earning and losing of points, but the game always ended with the particip ant having gained a positive value of points, which was meant to serve as an analogous p ositive outcome to that which those in the Cooperative and Individual conditions experi enced. Participants were further discouraged from attempti ng to keep track of their own points through being told that different question t ypes were assigned different point values. Participants completed the PANAS, predisposition ta sk, context manipulation, and Ring Measure. Measures of response latency were taken in the Ring Measure and predisposition tasks to confirm the strengths of in dividual decisions by participant. Instructions for each task were presented on-screen before the task, with the option for explanation from the researcher. Afterwards, partic ipants completed a manipulation check and were debriefed as to the nature of the de ception. Results In the predisposition (SVO pre-test) task, the part icipants arctangent values averaged by condition varied widely: in the individ ual condition, M =58.61, SD = 32.10;in the Competitive condition, M = 27.33, SD = 32.65; and in the Cooperative condition, M = 46.48, SD = 22.82. In the Ring Measure of Social Value Orientation (SV O post-test) task, the participants arctangent values averaged by conditi on were fairly similar: in the Individual condition, M = 2.10, SD = 0.85; in the Competitive condition, M = 0.72, SD = 1.82; and in the Cooperative condition, M = 1.60, SD = 1.03. (See Figure 6 for a bar chart
Social Decision Making in a Computer-Based Context | 26 of the means from this and the predisposition task, averaged by condition; see Figure 7 for a comparison of the mean SVO by group in the pr eand post-tests.) Figure 6 SVO angles from x-axis, by condition
Social Decision Making in a Computer-Based Context | 27 Figure 7 SVO preand post-test mean angles from the x-axis, by condition
Social Decision Making in a Computer-Based Context | 28 Discussion Computer-Based Context and Social Value Orientation The hypothesis that social value orientation after the context manipulation would correspond to the context in which participants par ticipated (Individualistic, Competitive, or Cooperative) was not supported by the data. Prio r to the context manipulation, the means of the SVO arctangent values by condition wer e very diverse, ranging across approximately 30 degrees from a low Pro-Social grou p orientation, to a Pro-social orientation which verged on being Altruistic. While this finding was odd prior to the manipulation, it made the collapse in arctangent ra nge after the context manipulation all the more apparent, when the outcome from the depend ent measure showed all mean orientations by group falling within 1.5 degrees of each other. Nearly all participants aligned with an Individualistic social value orie ntation after the completion of the context manipulation task, suggesting that completi ng a game-like task in a computerbased environment encouraged participants to behave individualistically after the completion of the task.This result did not follow f rom the literature about SVO decisionmaking priming in real-life scenarios, which typica lly shows that people can be verbally primed and take context into consideration when mak ing the decision to compete or cooperate (Drouvelis et al., 2010; Eiser&Bahvnani, 1974; Hertel& Fielder, 1994; Hertel& Fielder, 1998; Tremayne& Curtis, 2006; Zineldin, 20 04). These results suggest that computer-based contexts do not have a differential effect on decision-making which persists beyond the current task; but instead, that something about the task being computer-based draws out individualistic tendencies in people. The purpose of this experiment was to deter mine whether computer-based
Social Decision Making in a Computer-Based Context | 29 contextual priming in the completion of a task has an effect on decision making which persists after the task has been completed. In aski ng this question, the current study was conceptually replicating a study by Hertel& Fielder (1988), detailed earlier, which found that verbally priming the concepts of cooperation o r competition to be positivelyor negativelyvalenced can have an effect on decision -making outside of the tasks presented in the study, as measured by the Ring Measure: when for example, the concept of cooperation was primed with positively-valenced wor ds, participants were more likely to make cooperative decisions in later tasks. Though t he current study used the same pretest/post-test measures that were used in Hertel an d Fielder (1988), there are a number of differences between the two studies. The first majo r difference is the type of priming which the current study utilizes, that is, contextu al rather than verbal. This contextual priming was utilized to approximate the real-life e xperiences people might have outside of the lab setting while interacting with others th rough a computer. The findings from the Derks et al. (2004) study indicated that participan ts had the ability to distinguish between computer-based contexts. Those results were not sup ported in the current study. A possible explanation for this discrepancy is that t he types of context used in the current study were similar to each other in terms of their valence and task-oriented nature, unlike the types of context used in the Derks et al. study which varied between being taskoriented and socio-emotional in nature. Thus, it is possible that it was not as easy for participants in the current study to differentiate between these contexts within the greater context of completing a task on the computer. The other differences between these two studies, such as a lack of facial or verbal cuing i nvolved in the interactions between participants in the current study, and thus a reduc ed level of salience for the social nature
Social Decision Making in a Computer-Based Context | 30 of the task in the Competitive and Cooperative cond ition, are also found when comparing, more broadly, face-to-face and computerbased interaction. The current study attempted to increase this level of social salience through a number of means: firstly, participants were scheduled for data collection sim ultaneously with other participants whenever possible, allowing participants to physica lly see or even superficially interact with other participants before the task; secondly, in the verbal instructions that participants received, the existence of other playe rs in the context manipulation task was indicated; thirdly, participants were given a Play er Number, which they were instructed to enter before the context manipulation in order to enter the game,; finally, participants received feedback concerning their and the other players progress in the form of animated progress bars, as described earlie r. Given this focus on making the social nature of the tasks salient in this study, i t seems likely that the lack of social salience is inherent in the actual experience of in teracting via computer mediated communication (CMC), making these findings particul arly relevant to the question that the study addresses. Face to Face versus Computer Mediated Communication One of the hallmarkdifferences between CMC and F2F isthat it is frequently not possible to take advantage of such communicative to ols as facial and vocal cues between communicating parties in CMC. (This communicative d eficit is sometimes accommodated through the use of emoticons (Derks et al., 2004).)Additionally, participants seem to have difficulty differentiatin g between positivelyand negativelyvalencedsocial contexts in CMC. Hence,people behave individualistically. This result might not be surprising to users of the internet: a much-bemoaned side effect of web-
Social Decision Making in a Computer-Based Context | 31 based discourse is the tendency to behave rudely to ward or discount the opinions of others.The ways in which we communicate, interact, make decisions, and gather information online are likely impacted by these ind ividualistic tendencies. Implications in Web-Based Crowdsourcing Crowdsourcing is a means of outsourcing tasks or p rojects to a group of people. These tasks may be as diverse in purpose as large-s cale artistic collaborations, petitioning, information distribution, polling, or problem-solving (as in a software fix). The internet has recently become a convenient venue for crowdsourcing to take place, because it makes polls and collaboration between mi llions of people possible. Future iterations of this study might consider how individ ualistic behavior manifests itself on the internet, especially when people are asked to give an opinion or collaborate on a task: how are these tasks viewed by those partaking? Are people approaching collaboration differently online than they might in person? Are t he motivations the same? One might compose a study wherein performance on group collab orative tasks is compared in a computer-based setting and a non-computer based set ting, varying whether people are able to interact with their collaborators. The resu lts from the current study suggest that in such an experiment, people who are interacting thro ugh the computer might behave more individualistically than those who have the most aw areness of their group-mates, that is, those who interact with real people rather than oth er participants in a computer-based setting might be more inclined to make decisions le ss individualistically than those who are interacting solely in a computer-based setting as in this study.
Social Decision Making in a Computer-Based Context | 32 Implications in Web-Based Education These results have implications for another recent ly popular online collaborative venue: web-based education. As class crowding and accountability in education are on the rise, policy-makers have been searching for a s olution to the problem of providing children with more options for education. In attemp ting to provide the same level of education to all children, school districts have be en turning to web-based education. Increasingly, children are being compelled to take online courses to supplement or replace their face-to-face education, with some edu cational programs in Florida actually requiring multiple online courses as part of their curriculum track. As this form of education is on the rise, it seems topical to consi der the implications of attempting to replicate a classroom experience in a web-based are na through features which foster classmate interaction. In the State of Florida, the Florida Virtual Schools (FLVS) program is a state-backed online classroom which se rves this function, especially in crowded school districts. FLVS includes a number of features beyond the basic coursework, such as the ability to interact with on es classmates through forums or live chat, and the ability to chat with ones teachers. However, based on the failed efforts made in the current study to induce social salience in similar ways, such as giving feedback on the performance of the participant and other players and scheduling participants simultaneously such that they were abl e to see other people who were about to engage in the same task, these features may not be enough to truly induce pro-socially motivated collaboration or healthy competition in users. For better or for worse, removing the social or communal element from the ed ucation system may have consequences which might include reduced social mot ivation to perform (e.g., in a small
Social Decision Making in a Computer-Based Context | 33 group collaboration setting), reduced empathy for o thers (helping people by posting responses online versus helping someone sitting nex t to you), and possibly less pressure to perform even individually, as social comparison is less salient (Williams et al., 2000). It might be interesting, as a point of future study to compare the social value orientation of students engaged in a task in a traditional clas sroom setting versus in an online setting, to see whether this trend toward individualism in a computer-based context continues into web-based education. Conclusion Though the results of this study did not support th e hypothesis that context in a computer-based environment would affect the social value orientation-aligned decisions that participants made immediately following the ta sk, the fact that nearly all of the participants ended up with an individualistic align ment after being immersed in ANY computer-based condition after the task has interes ting implications for CMC. This tendency toward individualism in computer-based com munication should be kept in mind when considering web-based education policy and the results of crowdsourcing endeavors.
Social Decision Making in a Computer-Based Context | 34 Appendix A Items from the Positive And Negative Affect Scale ( PANAS-X)
Social Decision Making in a Computer-Based Context | 35
Social Decision Making in a Computer-Based Context | 36
Social Decision Making in a Computer-Based Context | 37
Social Decision Making in a Computer-Based Context | 38
Social Decision Making in a Computer-Based Context | 39
Social Decision Making in a Computer-Based Context | 40 Appendix B Social Value Orientation Predisposition Measure
Social Decision Making in a Computer-Based Context | 41 Appendix C Context Manipulation Items
Social Decision Making in a Computer-Based Context | 42
Social Decision Making in a Computer-Based Context | 43
Social Decision Making in a Computer-Based Context | 44 Appendix D Ring Measure of Social Value Orientation Items
Social Decision Making in a Computer-Based Context | 45
Social Decision Making in a Computer-Based Context | 46
Social Decision Making in a Computer-Based Context | 47
Social Decision Making in a Computer-Based Context | 48
Social Decision Making in a Computer-Based Context | 49 Appendix E Manipulation Check and Demographic Information Thanks for your time. I just have a few more questi ons for you. Gender: AOC: Age: Did you hear anything about this study prior to tak ing it/what/where? Did you notice anything unusual about the experimen t and what we told you about it? What do you think the purpose of the game you playe d (with the other players/against the other players) was? Please rate the extent to which you felt you were c ooperating in the game. 0 not at all 1 very slightly 2 a little 3 moderately 4 quite a bit 5 extremely Please rate the extent to which you felt you were c ompeting in the game. 0 not at all 1 very slightly 2 a little 3 moderately 4 quite a bit 5 extremely
Social Decision Making in a Computer-Based Context | 50 References Derks, D., Bos, A.E.R., & Grumbkow, J. (2004). Emot icons and social interaction on the internet: The importance of social context. Computers in Human Behavior, 23:1, 842-849. Drouvelis, M., Metcalfe, R., & Powdthavee, N. (2010 ). Priming cooperation in social dilemma games. Institut Zunkunft der Arbeit. 1-41. Eiser, J.R. & Bhavnani, K (1974). The effect of sit uational meaning on the behavior of subjects in the Prisoners Dilemma game European Journal of Social Psychology, 4, 93-97. Englemann, T., Tergan, S., & Hesse, F.W. (2010) Evo king knowledge and information awareness for enhancing computer-suppor ted collaborative problem solving. The Journal of Experimental Education, 78, 1-24. Frostling-Henningsson, M. (2009). First-person shoo ter games as a way of connecting to people: Brothers in blood. Cyberpsychology and Behavior, 12:5, 1-7. Griesinger, D.W., & Livingston, J.W. (1973). Toward a model of interpersonal motivation in experimental games. Behavioral Science, 18:3, 173-188. Hertel, G. & Fielder, K. (1994). Affective and cogn itive influences in a social dilemma game. European Journal of Social Psychology, 24, 131-145. Hertel, G. & Fielder, K. (1998). Fair and dependent versus egoistic and free: Effects of semantic and evaluative priming on the Ring Measure of Social Values. European Journal of Social Psychology, 28, 49-79.
Social Decision Making in a Computer-Based Context | 51 Komorita, S.S., Sweeney, J., & Kravitz, D.A. (1980) Cooperative choice in the N-person dilemma situation. Journal of Personality and Social Psychology, 38, 504-516. Liebrand, W.B.G. (1984). The effect of social motiv es, communication and group size on behavior in an N-person multi-stage mixed-m otive game. European Journal of Social Psychology, 14, 239-264. Liebrand, W.B.G., & McClintock, C.G. (1988). The ri ng measure of social values: A computerized procedure for assessing indi vidual differences in information processing and social value orientation European Journal of Personality, 2, 217-230. Loch, C. H., Galunic, D.C., & Schneider, S. (2006) Balancing cooperation and competition in human groups: The role of emotional algorithms and evolution. Managerial and Decision Economics, 27, 217-233. Murphy, R.O., Ackermann, K.A., & Handgraaf, M.J.J. (2010) Measuring social value orientation. Priutt, D.G. (1967). Reward structure and cooperati on: The decomposed Prisoners Dilemma game. Journal of Personality and Social Psychology, 7:1, 21-27. Tremayne, K. & Curtis, G.J. (2006) Cooperative beha vior in a Prisoners Dilemma a Terror Management perspective. 1-9. Zineldin, M. (2004) Co-opetition: The organization of the future. Marketing Intelligence & Planning, 22:7, 780-789.