Research and theory about the self developed over many years, largely independently of research and theory about groups. When theorists occasionally would seek to merge group theory and self theory, the focus was generally on the group self, as in shared group identity. In this manuscript, we make a case for the value of differentiated selves: Groups benefit greatly from differentiation of selves. The emergence of human selfhood might have been shaped by selective adaptation for playing an individual role in a group.
Allport (Reference Allport1924) wrote, “There is no psychology of groups that is not essentially and entirely a psychology of individuals” (p. 4). The point of departure for this manuscript is that Allport's assertion is fundamentally, even outrageously, wrong. In our view, the relationships among individuals are not fully reducible to properties of the separate individuals. On this, we think we are in good company (see Asch Reference Asch1952; Lewin Reference Lewin, Swanson, Newcomb and Hartley1952; Mead Reference Mead1934; Sherif Reference Sherif1936). Economic marketplaces cannot be reduced to the acts and choices of individuals; they comprise complex interactive systems. Much of social psychology's long tradition of research on groups has emphasized that groups are different from and more than the mere aggregate of their individual members. Selves thus do not constitute the group but rather play roles within the group's system. And differentiated roles make more powerful and effective systems.
A careful reading of the literature on groups yields not one but two thematic traditions denying that a group is equal to the sum of its parts. Unfortunately, their themes contradict each other. One line of work, dating back at least to Le Bon's (Reference Le Bon1896/1960) depiction of the group mind, depicts groups as generally worse than individuals acting alone. The other, whose exponents include the seminally influential economist Adam Smith (Reference Smith1776/1991), extols how groups produce and achieve far more than collections of independent, isolated individuals ever could.
The tension between these two traditions was apparent in two of the earliest works in social psychology. Triplett (Reference Triplett1898) observed and then confirmed empirically that people performed better in a group than when alone, in such domains as racing bicycles and winding fishing rods. Not much later, Ringelmann (Reference Ringelmann1913b; see also Kravitz & Martin Reference Kravitz and Martin1986) observed and confirmed empirically that people performed worse in groups than when alone, such as when pulling a heavy load together. Research in the modern era has continued to yield findings of both sorts, namely that being in groups sometimes makes people work harder and perform better but sometimes makes them slack off and perform worse than when alone.
In this manuscript, we propose, first, that both traditions of group research have valid points and important findings. It is quite true that sometimes groups are better than the sum or average of their parts – and in other cases they are far worse. Second, we shall propose the hypothesis that the difference can be explained largely on the basis of differentiation of individual selves. That is, groups surpass individuals when members of the group are individually identified and responsible, and when they contribute as distinct entities. Meanwhile, the worst outcomes of group processes come when individual identities are submerged in the group. By submerged in the group, we mean any of the following: People are held neither accountable nor responsible, they are not in competition or playing a distinct role, and they are not publicly identified or rewarded. It is a loss of individual or collective awareness of how group members differ from each other. Submersion of the self into the group is thus the opposite of differentiation.
An exhaustive review of all relevant work may be impossible – and certainly is impossible within the length constraints of journals such as this. Hence, our review is admittedly incomplete and selective. We reiterate that we seek to make the case for a theoretical position, and we welcome commentaries that provide alternate theories and additional evidence.
1. Theory: Why groups differentiate selves
People generally live in interacting groups, and they have done so everywhere on earth and throughout history. Groups confer benefits to individuals and can accomplish things that loners cannot. Groups also extract sacrifices. Group systems require individuals to set aside some self-interest, but members are tempted to pursue self-interest at group expense. Hence, group benefit depends on overcoming selfish desires so that people cooperate and contribute rather than free-ride or cheat. To be sure, the motivations of individual group members may vary from prosocial to selfish, as well as from eager for information to indifferent (De Dreu et al. Reference De Dreu, Nijstad and van Knippenberg2008). Managing the diverse and sometimes problematic motivations of individual members is often key to a group's success.
Two classes of reasons beyond self-interest will motivate people to contribute to group welfare, even at cost to themselves. First, if they love the group or identify passionately with it, they will want to advance its welfare and derive satisfaction from doing so. Second, they may contribute because the other group members put pressure on them to do so, such as by material incentives (e.g., rewards, punishments) and social incentives (e.g., moral reputation, laws). The second set of reasons thus reverts to appealing to the individual's self-interest and aligning it with progroup, prosocial behavior.
Some readers may regard the distinction between group goals and individual goals as artificial because groups consist of individual members and cannot really have motivations except in the minds of its members. Discussion of group goals is shorthand for saying that individual goals, right down to survival and reproduction, are facilitated by participation in groups, but achieving the benefits of groups often requires efforts, contributions, and sacrifices by individual members. Maximum individual advantage can be attained by sharing in group benefits without contributing, but if all members follow that strategy, there will be nothing to share. Groups therefore confer their advantages (and prevail over rival groups, thus also benefiting members) insofar as they motivate people to contribute, even to the short-term detriment of individual selfish goals.
1.1. Two complementary steps
The emergence of group activity can be divided heuristically into two steps. The first step involves the simple advantages of being in a group rather than alone. Belonging to the group is sufficient to furnish benefits that include collective vigilance, sharing of resources and information, and competitive advantages. Cohesiveness is a high priority for the group because it keeps members loyal and motivated to work with the group. The individual's goal is acceptance. Differentiation is not as important as shared identity at this stage.
The second step, our main focus, involves role differentiation. Role differentiation creates advantages and opportunities. It is no accident that all large corporations, governments, sports teams, and other such groups rely on it extensively. Larger groups permit more differentiated systems. Although animal groups may have some role differentiation, animal sociality does not have organizations with anything approaching the differentiated specializations found in a large (or indeed even a small) corporation or university.
Our account of these heuristic steps bears some resemblance to Tuckman's (Reference Tuckman1965) theory of group formation. He proposed that all groups require a “forming” stage, where acceptance and agreement are emphasized. This is followed by a “storming” stage, emphasizing differences and disagreements.
One key difference between the steps is whether the group functions mainly on the basis of how the various members are the same versus different. The benefits of shared group identity have been the focus of much theory and research, especially under the aegis of social identity theory (e.g., Hogg et al. Reference Hogg, Abrams, Otten and Hinkle2004; Turner & Tajfel Reference Turner and Tajfel1982). We seek to complement that work with an elucidation of the benefits of differentiation. Differentiation in this sense involves being individually identified and/or performing a distinct role as part of a system. Indeed, the effectiveness of the system may be based on different selves playing different roles. Differentiation should facilitate the gains drawn from systems as well as moral control of individuals by the group.
Role differentiation is thus not merely difference for the sake of difference but rather difference for the sake of facilitating systems. We use the term system gain to refer to the margin by which the members of a systematically organized group can achieve better results than the same number of individuals working together but without a system. A group may consist of various talented individuals who come together to compete against others in a battle, marketplace, or sports arena. That same group would be more likely to succeed, however, if they adopted a system that fosters performing complementary roles. The difference is system gain.
The crucial point is that system gain depends on differentiated selves. System gain capitalizes on members performing different roles. Specialization increases efficiency (individuals gain skill at their specific tasks and do not have to learn or perform other tasks) and quality (everything is done by an expert) (Smith Reference Smith1776/1991). In contrast, if everyone is the same and does the same things, that is hardly a system, and there will be no system gain. Differentiation underlies many features of groups that will figure in our literature review, including accountability and evaluation, responsibility, indispensability, and independent judgment.
The second step thus builds on the first. Although both steps (cohesive identification and differentiation) can occur at any point, we think there would generally be a sequence. The benefits of a cohesive group may occur quite early in group formation. Passionate commitment to the group (the first step) may motivate people to do their best in the short run, but in the long run it will almost certainly be useful for the group to hold individuals responsible for their actions, and so differentiation is needed.
Crucially, the individual's goals change at the second step. Merely securing acceptance is no longer sufficient. Being similar to everyone else and being a moral actor are key to the first step (gaining acceptance), but performance of individual, differential roles is key to the second. Hence, being different may become an important strategy in service of belongingness: A group cannot afford to lose a member who performs a unique function for the group, and so acquiring a unique skill can make someone indispensable. Being liked may be sufficient for the first step (gaining approval), whereas earning respect (by competent, ethical performance) becomes important at the second step.
The assertion that people have both a motivation to be different and a motivation to be the same as others in the group is the centerpiece of Optimal Distinctiveness Theory (Brewer Reference Brewer1991; Reference Brewer, Van Lange, Kruglanski and Higgins2012). The present approach acknowledges its debt to that theory and proposes one substantial change. In Optimal Distinctiveness Theory, the differentiation motive is postulated as something that requires no further explanation but is also linked to not being included in the group (Brewer Reference Brewer1991, p. 477). The implication is that people want to be close but not too close to others, and so they increase or decrease their conformity to gain acceptance or gain distance, respectively. In an important sense, then, the motive to differentiate is treated in that theory as going against the need to belong. This view has been preserved in many other influential theories about group processes (e.g., De Dreu et al. Reference De Dreu, Nijstad and van Knippenberg2008; Hinsz et al. Reference Hinsz, Tindale and Vollrath1997). In contrast, we regard differential individuation as a strategy to promote belongingness.
1.2. Forestalling potential misunderstandings
It is useful to distinguish two main kinds of group tasks: productive achievement and information use (e.g., sharing and accumulating knowledge, group decision-making). Both can benefit from differentiated selves, but the role of group control is different. When productive achievement costs effort or other resources, groups benefit from public differentiation, which lets them monitor individual efforts and hold people responsible, such as by rewarding high contributors and punishing slackers and cheaters (Leary & Forsyth Reference Leary, Forsyth and Hendrick1987). For informational tasks, private differentiation benefits the group by promoting individual thought and judgment, whereas group control promotes conformity and undermines independent thinking.
Anonymity is thus not the opposite of differentiation and at times can even facilitate it. Anonymity protects individuals from being controlled by the group. For informational tasks, such as voting, anonymity can help ensure independence of judgment. However, in performance contexts, anonymity may detract from good group outcomes by protecting free riding and other selfish, antisocial acts.
Selfishness is not the same as human selfhood and in fact long precedes it. Selfishness is rooted in the very nature of life, insofar as every living organism delineates a boundary between itself and its environment. It lives or dies as a totality, and its motivations are designed by natural selection to promote and prolong its life (plus kin and offspring). Human selves have this same selfish core – alongside additional features that enable them to overcome this natural selfishness if there are good reasons to do so. The desire to achieve social acceptance within a group may provide just such a reason.
1.3. Moral control in large groups
Groups benefit insofar as individuals follow the rules and do what is best for the group – that is, groups benefit from moral behavior. The first step, identifying with the group, can motivate people to do what is good for the group, and so it can yield some improvement in moral behavior. With the second step, however, the group can exert control over individuals by holding them accountable. Thus, the first step relies on inspiration and voluntary self-sacrifice to improve moral quality, which can be effective at times, especially when there are strong emotional bonds. The second step enforces moral behavior by rewarding virtue and punishing vice, and so in the long run it is likely more effective than the first at promoting moral behavior. The sequence is evident in macrosocial trends. Friedman (Reference Friedman2002) pointed out that moral rules and laws generally promote quite similar behaviors, but the motivational basis changes as societies evolve. In small groups characterized by stable relationships based on emotional bonds, people care about each other and reputation, and so people are motivated to act morally. As society grows larger and interactions with strangers increase, the (weaker) emotional ties become inadequate to ensure good behavior, and so moral suasion is replaced by law enforcement.
We assume competition among groups has been an important factor in human evolution. Successful competition depends on size and system. In many competitions, larger groups tend to prevail. Primitive battles were generally won by the larger group (e.g., Morris Reference Morris1965), and achieving numerical superiority has been a major goal of modern military efforts, too. In fact, many major wars have ended with twice or three times as many soldiers under arms as began the war (Hubbard & Kane Reference Hubbard and Kane2013) – despite extensive casualties.
As groups became larger and more evenly matched, a second factor, role differentiation, provided powerful advantages (e.g., McNeill Reference McNeill1982). This is the crux of our model: Groups do best when they start by developing commitment and identification in a group of individuals and then move to instantiating and emphasizing distinct identities and roles, especially as the group gets larger.
An authoritative review by Levine and Moreland (Reference Levine and Moreland1990) concluded that most factors that make groups effective and satisfying deteriorate as group size increases. On that basis, one might anticipate that people would eschew large groups, whereas in empirical fact historical progress has seen gradual increases in operative group size. Large groups must thus have some compelling advantages – yet they also must become able to function without some of the motivational processes found in small groups. These advantages may derive from simple numerical advantage (e.g., more warriors on the battlefield), but many depend on differentiation. Large groups can provide much more differentiation and specialization than can small groups. Hence, large markets, large universities, and large corporations have advantages over smaller ones, especially in terms of greater specialization. Both informational and performance goals are served by having many individuals contributing their unique talents, knowledge, and expertise.
The emphasis on differentiation thus may come after initial drives for acceptance, partly because of the tendency for successful groups to grow larger and less intimate over time. It is well established that in large groups, feelings of social connection are weaker than in small groups (e.g., Levine & Moreland Reference Levine and Moreland1990; Mueller Reference Mueller2012). A larger group is therefore more likely to have slackers and other rule breakers, and so individual identification is useful for motivating people with rewards and punishments.
An example from ancient Chinese history helps illustrate our two-stage model. At one point, ten-thousand (!) independent political domains consolidated into seven. According to Fukuyama (Reference Fukuyama2011), this was accomplished mostly by larger groups conquering and integrating their smaller neighbors (so having many members was decisive for group success). The ensuing process by which the seven merged into one China was dominated by the complex administrative and military systems using extensive role differentiation developed by the Qin (so system gain was decisive).
2. Review of evidence
We turn now to a presentation of research findings. Our central hypothesis is that groups will produce better results if the members are individuated than if their selves blend into the group. We posited that being identified with and accepted into the group is essential in the initial stages. Self-sacrifice for the group's goals can come at this stage, following from commitment to the group. Later, the group will be successful to the extent that it fosters individuality.
The section first addresses the two main types of group process: group task performance (2.1) and then informational processes (2.2). Following this, two further sections examine the broader question of group moral control (2.3) and then evidence about the two-step sequence (2.4).
2.1. Task performance
2.1.1. Social facilitation
Social facilitation largely involves improvements in effort and performance caused by the presence of others. To be sure, sometimes the mere presence of others impairs performance, especially when complex, poorly learned tasks are involved (Zajonc Reference Zajonc1965). But such tasks mainly require skill, and there is little a group can do in the short run to improve skilled performance. Performance gains are presumably based on increased effort. These fit our theme that groups seek to control individuals and improve their performance by means of identifying them individually so as to incentivize effort.
Several factors amplify social facilitation: individual identification, accountability, and anticipated evaluation (which motivates the desire to be favorably regarded by others) (for reviews, see Bond & Titus Reference Bond and Titus1983; Geen & Gange Reference Geen and Gange1977; Guerin Reference Guerin1986). These are possible only based on identifying people individually.
Competition pits individuals against each other and thus invokes evaluation, accountability, and other hallmarks of individuality. VanTuinen and McNeel (Reference VanTuinen and McNeel1975) showed that performance improved with explicit competition but not when participants merely worked together. In another condition, performance improved based on a cash incentive despite working alone. Thus, improvement stemmed either from competing against someone or from working as an individual for a contingent reward. Both differentiate the self (to compete and to seek individual reward).
Many studies have provided evidence that one reason performance improves in the presence of others is that people want to perform well so others will think well of them (evaluation apprehension; Cohen & Davis Reference Cohen and Davis1973; Feinberg & Aiello Reference Feinberg and Aiello2006; Henchy & Glass Reference Henchy and Glass1968; Martens & Landers Reference Martens and Landers1972; Rajecki et al. Reference Rajecki, Ickes, Corcoran and Lenerz1977). Good (Reference Good1973) showed that performance improved when participants were told that the experimenter (as opposed to a computer) would evaluate their performance. In a further twist, Good found that only participants who had been led to expect a favorable evaluation showed the performance improvement, which suggests that expecting an unfavorable evaluation can wipe out the gains from evaluation apprehension (see also Bray & Sugarman Reference Bray and Sugarman1980; Green Reference Greene1979). (Thus, perhaps evaluation optimism rather than evaluation apprehension would be the more precise term for what causes performance to improve.)
Evaluation by others, rather than self-evaluation, appears to be crucial for social facilitation (Szymanski et al. Reference Szymanski, Garczynski and Harkins2000). Thus, social facilitation is about the self as seen by others. Bond (Reference Bond1982) showed that performance is not simply a matter of arousal and item difficulty – rather, it depends on the composite image of self that one thinks is being communicated in that situation. He showed that when easy items were embedded in a set of mostly difficult items, performance was impaired even on the easy ones. Conversely, when a few difficult items were embedded in a mostly easy problem set, performance was unimpaired.
The Köhler effect is the change in performance due to a person's awareness of its impact on others: The least-capable member of the group sometimes performs better in the group than he or she would if performing alone (Köhler Reference Köhler1926). Swimmers and track athletes often do better as part of relays than when alone, with gains found mainly among the weaker members (Hüffmeier & Hertel Reference Hüffmeier and Hertel2011; Osborn et al. Reference Osborn, Irwin, Skogsberg and Feltz2012).
To summarize: The idea that people perform better in groups is one of the fundamental arguments for the value of groups, and so social facilitation is likely adaptive. Crucially, however, these benefits occur mainly when people are individually identified and motivated to care about how others will evaluate them – consistent with the view that differentiated selfhood facilitates group function.
2.1.2. Social loafing
Social loafing is the tendency for people to reduce effort when in a group. The reduction of effort produces an overall loss of output, because the members of the group do poorer work and produce less than they would produce individually. Ringelmann (Reference Ringelmann1913b) first verified a drop in performance in the context of teams of men pulling together with less force than the sum of their individual efforts. With methodological refinements, Latané et al. (Reference Latané, Williams and Harkins1979) replicated the phenomenon: Participants worked harder when alone than when part of a group. Latané et al. found loafing even when people actually performed alone but merely believed they were part of a group. Williams et al. (Reference Williams, Harkins and Latané1981) showed that identifying people individually eliminated social loafing.
The theory of social loafing derived from earlier work on diffusion of responsibility, a pattern by which the pressure to take action is divided among the group members (Darley & Latané Reference Darley and Latané1968). Being anonymous, so that one's identity is submerged in the group, increases the tendency for everyone to leave difficult or risky jobs for someone else (e.g., Schwartz & Gottlieb Reference Schwartz and Gottlieb1976; Reference Schwartz and Gottlieb1980). The larger the group, therefore, the more diffusion of responsibility.
A meta-analytic review by Karau and Williams (Reference Karau and Williams1993) confirmed that social loafing is reduced by making people identifiable, especially when individual evaluation is possible. Also, social loafing is reduced by giving people nonredundant roles in the group (i.e., indispensability), so that one member's lack of contribution will not be offset by another member's performance. Karau and Williams proposed that social loafing depends on people's appraisal of how much the group performance depends on their own contribution, how much group outcomes depend on group performance, and how much the individual will benefit from the group's performance and outcomes. People work hardest when they believe their individual effort contributes to outcomes that benefit both the group and the self (also Karau & Williams Reference Karau and Williams1995). For example, Weldon and Mustari (Reference Weldon and Mustari1988) provided evidence that social loafing occurs mainly when people believe their contribution to the group is dispensable. Likewise, they found that feeling that one's own work is indispensable could motivate high effort and good work even when one is anonymous.
Evidence for the importance of moral control was provided by work on perceived procedural fairness (De Cremer et al. Reference De Cremer, Hoogervorst and Desmet2012). Leaders often punished the least-contributing member of the group. Such practices communicate to group members that their behavior is being individually tracked. Group performance improved as a result of this sort of legitimate individuation and punishment.
In sum, the social loafing literature confirms the general pattern that group performance is harmed when people feel submerged in the group and improved when group members are individually identified. It also shows that people perform well when they have unique roles and make contributions they regard as indispensable. These fit the main themes of our theory. Individual identification facilitates responsibility and accountability, thus putting moral pressure on individuals to behave well. Indispensability involves differentiating members' roles, which is useful for groups that have complex systems. There are certainly ample signs of the first step of group work: Caring about the group and identifying with it reduce social loafing. But differentiation improves the group's ability to motivate people to exert themselves on behalf of group goals.
2.1.3. Division of labor
Division of labor is one powerful process by which individual differentness can improve outcomes via system gain. Adam Smith's (Reference Smith1776/1991) classic treatise on economics began by discussing the benefits of division of labor in a pin factory. Division of labor enabled the factory to produce far more pins than a comparable number of individuals working separately. Babbage (Reference Babbage1832) explicated the monetary savings to an organization that stemmed from extending division of labor to subtasks. Artisans who built entire products needed a wide range of skills and hence were expensive. Dividing the task into segments, each performed by a different person with a narrow skill set, reduced labor costs considerably while also improving quality (because of specialization). Thus, seemingly paradoxically, a collection of individuals with relatively limited skills could outperform a collection of experts who did not specialize.
The benefits of division of labor are now widely accepted. Without it no large organization could be successful. West (Reference West1999) compared flute manufacturing at two factories that were quite similar except that one used a 19-step division of labor whereas the other had no division of labor. Productivity was almost four times higher with than without division of labor. Another study with a Japanese bank found improved performance due to specialization, which is another aspect of division of labor (Staats & Gino Reference Staats and Gino2012).
Although division of labor is mainly a topic for other disciplines, such as economics and sociology, psychology has made some useful contributions. Research on transactive memory has found that groups remember things better insofar as they assign various members to specialize in remembering different things (Wegner Reference Wegner, Mullen and Goethals1986). A meta-analysis, including studies of actual organizations and ongoing workgroups in business, sports, and military combat, found that performance was indeed substantially improved insofar as group members specialized in their knowledge and kept track of who knows what (DeChurch & Mesmer-Magnus Reference DeChurch and Mesmer-Magnus2010). Crucially, they found that the benefits of differentiation tended to emerge over time, which is consistent with our two-step model. Shared identity is crucial in the first stage and promotes quick gains, whereas differentiation improved performance over the long run.
That groups spontaneously seek to divide labor so as to maximize outcomes was suggested by Chatman et al. (Reference Chatman, Boisnier, Spataro, Anderson and Berdahl2008). They showed that when a group contained only one member of a particular gender, others deferred to that person on tasks relevant to that gender, and the person's performance improved. Thus, having an individual identity within the group that marks one out as likely to be good at something causes one be accorded something like expert status on that task, as reflected in being treated as a leader – and it improved performance of the differentiated person.
The benefits of division of labor arise not just from having different people do different things but also from coordinating them into an integrated system. Specialized expertise at making one part of a flute is useless unless there are other specialists who make the other parts. A group performance study (Stasser et al.Reference Stasser, Stewart and Wittenbaum1995) used a hidden profile mystery task in which the solution was known only when the group integrated information possessed by different members. The best performance came from groups in which members knew which other members had which kinds of relevant information. The researchers concluded that cognitive division of labor requires group members to know who knows what. This is relevant to our theme that the benefits of the group require that members know each other's differentiated identities within the group.
A field study at several call centers in India manipulated the initial training to emphasize either trainees' individuality and unique potential contribution, the greatness of the organization, or skills training (neutral control group) (Cable et al. Reference Cable, Gino and Staats2013). Individual differentiation yielded the best results, both in terms of performance (measured by customer satisfaction) and staying with the organization over six months. A laboratory study yielded similar results, with performance, retention (returning for a second session), and subjective engagement highest among those whose initial instructions had emphasized individual, unique contribution rather than emphasizing being part of a group that already does wonderful things. In this case, at least, being different was more motivating than being the same.
In sum, division of labor is one of the founding principles of economic organization and human group performance because it confers huge benefits on most task performance. Division of labor is a paradigmatic example of system gain, and it is essentially based on differentiated selves who perform distinct, yet complementary, tasks. Its benefits are especially pronounced when members know who will do what and trust each other to follow through.
2.2. Information, judgment, decision
We turn now to reviewing phenomena related to the informational function of groups. Thinking in groups differs from thinking by individuals, for example, in being simpler and more homogeneous (Hinsz et al. Reference Hinsz, Tindale and Vollrath1997). Le Bon's (Reference Le Bon1896/1960) characterization of the group mind as primitive and irrational provided an influential statement of the pessimistic view that groups are less intelligent than individuals. Optimism is, however, apparent in many quarters, not least in the endless proliferation of committees in all institutions.
2.2.1. Pooling information for group decisions
Work by Stasser and colleagues (e.g., Stasser & Titus Reference Stasser and Titus1985; Reference Stasser and Titus1987) cast doubt on the value of committees. Their studies used the hidden profile research design: Certain information is given to individual members, other information is given to the group as a whole. Stasser et al. gave a large amount of information favoring one position “privately” to many individual group members. For the group as a whole to accurately gauge the strength of this position, all members needed to reveal the information they possessed. On the other hand, less support for the contrasting position was provided, but this information was given “publicly” – that is, to the group as a whole. The primary rationale for having committees is that members can pool their various knowledge to produce a full picture (i.e., reveal the hidden profile). Unfortunately, the usual finding has been that group members talk about the information they all have in common, and the individually held bits of information get left out of the discussion and decision-making processes. Hence, committees make inferior decisions because they fail to capitalize on the differentiated knowledge of individual members (Stasser Reference Stasser, Levine, Thompson and Messick1999; Wittenbaum & Park Reference Wittenbaum and Park2001; Wittenbaum & Stasser Reference Wittenbaum, Stasser, Nye and Brower1996).
A meta-analysis by Lu et al. (Reference Lu, Yuan and McLeod2012) confirmed that these effects are quite large. Groups talked about shared information far more than individually held information. The larger the group, the more members focused on what information they had in common and the more that tendency degraded the quality of the decision. In short, hidden profiles generally remained hidden, especially in larger groups.
We have observed that groups often treat cohesion as a goal. Although cohesion may seem especially desirable when consensus is sought, it does not necessarily improve the quality of group decisions. If group members know there is dissent among them, they become more likely to bring up their unshared knowledge, improving the quality of group decision (Brodbeck et al. Reference Brodbeck, Kerschreiter, Mojzisch, Frey and Schulz-Hardt2002). More broadly, research on minority influence has shown that a persistent minority can stimulate the majority to think more carefully about the issue facing the group, thus ultimately improving the group (Nemeth Reference Nemeth1986). Disagreeing minorities may thus reduce cohesion (by undermining consensus), but in the long run they can facilitate better information and decision-making processes.
Indeed, the mere fact of dissent seems to improve decision quality, even if the dissent consists wholly of advocating different non-optimal options. Schulz-Hardt et al. (Reference Schulz-Hardt, Brodbeck, Mojzisch, Kerschreiter and Frey2006) showed the positive effect of dissent nicely, instructing three-person groups to make a hiring decision among four candidates. When group members started out arguing between two non-optimal candidates, they talked long enough to allow the hidden profile (favoring a third candidate) to emerge. Nemeth (Reference Nemeth1986) also found that dissenting minorities could benefit the group even if the majority were not won over to the minority's view because the majority would respond to dissent with divergent thinking and thereby might discover new facts and options. The value of differentiated selfhood is therefore not restricted to cases in which a minority advocates the best answer – differentiation helps even if no one initially advocates the best answer.
2.2.2. Brainstorming
The purpose of brainstorming is to generate creative ideas. The practice was first developed in advertising agencies (see Osborn Reference Osborn1953). In brainstorming, group members share insights and ideas, stimulating each other toward more creative output.
In general, the early enthusiastic reports of brainstorming's effectiveness were followed by a mass of sobering data that repeatedly found brainstorming groups produced fewer and lower-quality ideas than the same number of individuals working alone (for meta-analysis, see Mullen et al. Reference Mullen, Johnson and Salas1991). These results could be a result of social loafing or to reduced effort on the part of members who feel their contributions are not unique – or even that those contributions may be dispensable.
Early rules for brainstorming groups prohibited criticism. In practice, members did sometimes criticize each other, and this was regarded as a deterrent to productivity. Recent work, however, suggests that the performance of brainstorming groups actually improves when people criticize each other (Nemeth et al. Reference Nemeth, Personnaz, Personnaz and Goncalo2004). Thus, again, differentiating individual identities appears to improve the performance of groups.
2.2.3. Conformity
One finding that stimulated conformity research came from Asch's (Reference Asch1952) research, in which participants provided answers that were clearly false if those answers were given by all other members of the group (who were confederates of the researchers). A review by Bond and Smith (Reference Bond and Smith1996) upheld the basic finding and concluded that conformity is higher to the extent that people are emotionally invested in the group and wish to maintain cohesion. Deutsch and Gerard (Reference Deutsch and Gerard1955) showed that when pressure to conform was reduced by offering anonymity, members were more likely to express their individual opinions: Anonymous members therefore made fewer mistakes than identified ones. Being identified to someone outside the group (i.e., the experimenter) also improved accuracy. Increasing the pressure for conformity, however, heightened the tendency to agree with the group's false assertion. More recent work has confirmed that people who resist the tendency to conform to the group's opinion can improve the informational performance of groups (Madirolas & de Polavieja Reference Madirolas and de Polavieja2014).
The benefits of anonymity for judgment quality contrast with its costs in effortful performance (as the social loafing section showed). Making people anonymous rather than identifiable increased social loafing but improved their willingness to express novel opinions during group decision tasks. In both cases, however, the optimal result depends on getting the person to behave as an autonomous, independent, responsible individual. As we explained in the theory section, anonymity shields the individual from group control, which can facilitate laziness and free riding but also frees people to think and judge independently.
Although early work by Schachter (Reference Schachter1951) showed that groups tend to dislike and reject dissenters who espouse opinions at odds with the emerging consensus, dissenters can be extremely valuable in improving group decisions. Schachter (Reference Schachter1954) found that some groups even ended up agreeing with the dissenter. But groups often reject a dissenter despite his or her potential value. As we theorized, the first step in group formation involves harmonious relationships to integrate individuals, and the second step improves performance by means of differentiated roles. Dissent may detract from the first even while benefiting the second. Research on minority influence, in particular, has shown that a dissenting minority can improve the thinking of the majority, despite some negative reactions deriving from the initial loss of consensus (Nemeth Reference Nemeth1986). Moreover, the negative emotional and interpersonal reactions to dissenters constitute palpable pressure on people to conform – and many do, to the detriment of the group's ability to profit from the diverse perspectives and knowledge of its members. Groups even go as far as ejecting dissenters if given the chance (Schachter Reference Schachter1951; Tata et al. Reference Tata, Anthony, Lin, Newman, Tang, Millson and Sivakumar1996).
2.2.4. Groupthink
Janis's (Reference Janis1972) influential critique of group cognition and decision-making, under the rubric of groupthink, also highlighted the role of dissolving into the crowd. Janis showed how committees and other groups had made costly and seemingly avoidable errors when all members focused their thinking on the same assumptions and information. Having reviewed the literature, Esser (Reference Esser1998) remarked on the contrast between the hundreds of articles that cite groupthink and the relatively few direct empirical tests. Still, she concluded that the theory of groupthink had fared reasonably well empirically, although some factors, such as time pressure and group cohesion, had not played the vital roles the theory had suggested. The general implication is that group decision-making is improved by differentiation and impaired by uniformity.
Some of the relatively poor thinking of groups is likely produced by social loafing, diffusion of responsibility, and the consequent reduction of cognitive effort. Petty et al. (Reference Petty, Harkins and Williams1980) showed that participants put less effort into various evaluative tasks when they were working in groups of 10 or 15 than when alone. Informational input (e.g., quality of argument) had stronger effects on individuals than on groups, and individual evaluations were stronger and more extreme than group ones, again reflecting the pattern of people putting less effort into the decision when they are part of a large group.
2.2.5. Accountability
Accountability has been defined as the expectation that one may have to justify one's beliefs, feelings, and actions to others (Lerner & Tetlock Reference Lerner, Tetlock, Lerner and Tetlock2003; Scott & Lyman Reference Scott and Lyman1968; Semin & Manstead Reference Semin and Manstead1983; Tetlock Reference Tetlock and Zanna1992). Accountability emphasizes the responsibility of individuals to behave autonomously and present a valid basis for their actions, so it individuates group members. This can help the group succeed even when the motivations of individuals might be counterproductive, such as by being selfish or having low interest in the group's informational goals (De Dreu et al. Reference De Dreu, Nijstad and van Knippenberg2008).
Accountability can overcome some of the informational failures already covered, such as the committee effect. Scholten et al. (Reference Scholten, van Knippenberg, Nijstad and de Dreu2007) improved the quality and accuracy of group decisions by telling participants that they would have to explain the decision process later. De Dreu and van Knippenberg (Reference De Dreu and van Knippenberg2005) showed that process accountability (i.e., knowing that one would have to justify how one reached one's decisions) reduced the negative reactions to people who brought up contrary views.
Likewise, accountability can improve the otherwise poor performance of brainstorming groups. When participants expected to have to explain and justify the process of generating ideas, they generated more ideas than in the nonaccountable groups (Bechtoldt et al. Reference Bechtoldt, De Dreu, Nijstad and Choi2010). Indeed, one procedure that greatly improved the performance of brainstorming groups involved having participants first generate ideas individually and then bring them together to evaluate and combine them (Lamm & Trommsdorff Reference Lamm and Trommsdorff1973; Mullen et al. Reference Mullen, Johnson and Salas1991). Thus, individualizing the process improved group performance.
Accountability makes people think more thoroughly and carefully about their tasks than they would otherwise. This benefits the group by improving quality. Tetlock (Reference Tetlock1983) had participants simulate being jurors and form judgments about a defendant's guilt. An irrational (primacy) bias was eliminated by telling participants in advance they would have to explain and justify their decisions. Weldon and Gargano (Reference Weldon and Gargano1988) likewise found that accountability (expecting to have to explain one's ratings and decisions) reduced diffusion of responsibility and social loafing.
A review by Lerner and Tetlock (Reference Lerner and Tetlock1999) concluded that only some types of accountability increase mental effort, and moreover that increased effort is not necessarily beneficial. Being accountable to an audience or authority who values accuracy and fair process motivates people to try to be fair, objective, and accurate. But accountability to a biased authority or audience who desires a particular conclusion can increase bias toward that conclusion (Tetlock et al. Reference Tetlock, Skitka and Boettger1989). Accountability may also increase bias when the biased option is easiest to justify to others. Subsequent work found that sometimes people react to accountability with evasive tactics and buck-passing, so as not to be blamed for problematic stances (Green et al. Reference Green, Visser and Tetlock2000). Still, in general its effects are beneficial more often than not.
Thus, the general pattern seems to be that accountability makes the person do what the group wants. This motivation is helpful when it leads to more careful and systematic thinking and therefore greater accuracy, but it is detrimental when it leads to embracing the group's biases. Admittedly, classifying those outcomes as helpful versus detrimental rests on assumptions that finding the truth is the supreme goal. Group cohesion and agreement may sometimes be higher priorities than the truth, and certainly many groups have been more interested in supporting their values and ideologies than in an open-minded quest for truth. Such groups might therefore regard accountability as helpful even in some cases that we have labeled detrimental.
2.2.6. Wise groups
Despite the accumulation of findings indicating collective stupidity, it is possible for groups to perform feats of remarkable intelligence. Surowiecki (Reference Surowiecki2004) presented multiple lines of evidence to indicate that the pooled knowledge of individuals can often outperform even experts. In one dramatic study, he compiled data from the television game show Who Wants to Be a Millionaire?, on which stumped contestants can consult various helpers. Contestants who asked their favorite expert did fairly well, getting the question right 65% of the time. Surprisingly, however, those who polled the studio audience did better, with a remarkable 91% correct.
How can crowds of individuals outperform knowledgeable experts? Surowiecki (Reference Surowiecki2004) concluded that collective wisdom arises from highly individualized judgments: People make their own choices, largely independent of what everyone else thinks. For example, sports bettors win or lose money based on their individual bets. Groupthink and conformity pressures are minimal and hence unable to influence how an individual votes. Random errors will cancel each other out in a large sample, but if people make similar errors (because of, for example, bias or common intuitive processes) then accuracy will be reduced (Simmons et al. Reference Simmons, Nelson, Galak and Frederick2011).
The wisdom of crowds is also, clearly, the principle underlying the usefulness of democratic voting by secret ballot. Some evidence has confirmed the benefits of secret voting. Two investigations used random assignment to condition in order to engineer how inhabitants of 299 villages in Afghanistan and Indonesia made decisions on which projects they wanted to pursue as part of a program funded by international nongovernmental organizations. In half of the villages, elites or other representatives made the decisions, whereas in the other half, villagers voted by secret ballot to decide which projects to pursue. The villages were generally unfamiliar with secret ballots, whereas decision by elites had the advantages of tradition and familiarity. Yet large, robust findings indicated that the secret ballot yielded better outcomes, including objectively superior choices, greater satisfaction, and more perceived benefits among the villagers even a year later (Beath et al. Reference Beath, Christia and Enikolopov2012; Olken Reference Olken2010).
Comparison of voting records of people who do versus do not believe that their votes are secret show that those beliefs have effects (Gerber et al. Reference Gerber, Huber, Doherty and Dowling2013). Labor union members who doubted the secrecy of their votes were less likely to vote against the union's preferred candidates than those who believed their votes were safely confidential.
2.2.7. Conclusion
The intelligence of groups has been much discussed and debated, and replicable examples of both collective wisdom and collective stupidity have been found. The positive outcomes reflecting intelligent, wise decisions, and good, creative problem-solving performance are generally associated with people acting as independent selves, whereas submersion of individual selves in the group produces the negative outcomes. Expecting to be evaluated individually (accountability) and performance of unique, independent roles in the group tend to produce the best results. Thus, again, groups benefit from the autonomous operation of individual selves.
2.3. Prosocial and antisocial behavior
Thus far, we argued that group task performance and information management are both facilitated by differentiating selves, but the difference depends on implications of group moral control of the individual. Lack of identification frees individuals to misbehave by slacking off on effort tasks but frees them to think individually on information tasks. If public identification facilitates groups' moral control (good for effort management, bad for information and judgment diversity), it should generally push toward more prosocial than antisocial choices. For example, anonymous donations to charity are vastly smaller than identified ones (Satow Reference Satow1975).
2.3.1. Commons dilemma and other social dilemmas
Hardin (Reference Hardin1968) invoked the “tragedy of the commons” to explain the destructive depletion of commonly held grazing areas. When individuals are responsible for their land and livestock, they maintain their herd and land so that the grass continues to grow back, thereby making the resource sustainable – but when the land is held in common, individuals grow their herd and let it consume freely until the resource is overused to the point that it fails to renew.
Many studies have shown that identification and accountability can improve outcomes in the commons dilemma and similar situations. For example, De Kwaadsteniet et al. (Reference de Kwaadsteniet, van Dijk, Wit, De Cremer and de Rooij2007) manipulated accountability by telling people that other group members would know how much they took from a renewable common resource pool and by telling them they would have to justify their actions later. Accountability improved the sustainability of the resource and thereby improved the entire group's long-term outcomes. Several studies have shown that the larger the group, the less cooperation and restraint members show, presumably because large groups increase diffusion of responsibility (Messick & Brewer Reference Messick, Brewer, Wheeler and Shaver1983; Orbell & Dawes Reference Orbell, Dawes, Stephenson and Davis1981).
With resource-contribution games, selfishness prescribes not contributing whereas the group benefits if everyone contributes. Cabrera and Cabrera (Reference Cabrera and Cabrera2002) concluded from multiple studies that publicly recognizing people's individual contributions increased the total contributed. One important aspect of the commons dilemma is that people expect the resource to be depleted despite any restraint on their own part; if one does not take extra resources someone else inevitably will. In other social dilemmas, too, the belief that one's own efforts or contributions can be replaced by others may contribute to making individuals behave selfishly.
There are at least two ways to break this destructive cycle of self-fulfilling expectations of mutual failure. One is to foster the belief among members that they can count on each other to serve the greater good rather than narrow self-interest. De Cremer et al. (Reference De Cremer, Snyder and Dewitte2001) showed that people's willingness to restrain themselves and help the group depended substantially on whether they could trust others to do likewise. Both steps in the model are relevant: People may trust others to contribute because the group members share feelings of solidarity and commitment or because members are accountable and free riders can be found out and punished.
The other antidote to destructive expectations (i.e., that one's lack of contribution will not matter because others will compensate) is to structure the situation so that each person's contribution is indispensable in some way. This pertains to the second step in our model, which highlights the importance of differentiation. Multiple investigations have shown that making individual contributions indispensable can help solve social dilemmas (Kerr & Bruun Reference Kerr and Bruun1983; Lynn & Oldenquist Reference Lynn and Oldenquist1986; Stroebe & Frey Reference Stroebe and Frey1982; Van de Kragt et al. Reference Van de Kragt, Orbell, Dawes, Braver, Wilson, Wilkc, Messick and Rutte1986). Indispensability obviously depends on differentiation: One's role in the group is not unique insofar as one's contribution can easily be replaced by other members.
Indeed, accountability improves prosocial behavior in social dilemmas. People contribute more to the group resource pool if there is a system for punishing free riders (Fehr & Gächter Reference Fehr and Gächter2002). But such systems are costly to maintain because members must make sacrifices to enforce punishment. De Cremer and Van Dijk (Reference De Cremer and Van Dijk2009) showed that people make more such contributions if they expect to have to justify their actions, as compared to no accountability.
As for trust, research in accounting has suggested that individualized record keeping can enhance it and thereby facilitate system gain. Basu et al. (Reference Basu, Dickhaut, Hecht, Towry and Waymire2009) conducted an experiment with the economic trust game to show that in complex environments, allowing people to keep records of everyone's prior actions increased trust and increased the total yield on investment, thereby enriching everyone. Record keeping enabled mutually beneficial exchanges to increase, whereas defection and exploitation were penalized, thereby improving the moral quality of the social group. The authors extrapolated from their findings to note that in human history, the advent of record keeping (which depends on individual identification and accountability) enabled substantial gains in trade, wealth, and morality.
Some findings indicate that people sometimes do things to benefit the group as a result of feeling personally identified with the group. These findings point to the first step in group formation (developing a common bond), the precursor to our emphasis on role differentiation. The more that members identify with the group, the more they contribute to public goods games (De Cremer & van Dijk Reference De Cremer and Van Dijk2002; De Cremer et al. Reference De Cremer, Van Knippenberg, Van Dijk and Van Leeuwen2008). Kramer and Brewer (Reference Kramer and Brewer1984) found that people sustained the resource longer in a commons dilemma game if their collective social identity was made salient (see also Goldstein et al. Reference Goldstein, Cialdini and Griskevicius2008; Tyler & Degoey Reference Tyler and Degoey1995).
Various other studies have also shown improved cooperation in commons dilemma and other social dilemma situations as a result of enhancing a sense of group identity (Brewer & Kramer Reference Brewer and Kramer1986; Dawes et al. Reference Dawes, Van de Kragt and Orbell1988; Rapoport et al. Reference Rapoport, Bornstein and Erev1989). By way of explanation, Van Lange et al. (Reference Van Lange, Liebrand, Messick, Wilke, Liebrand, Messick and Wilke1992) proposed that “group identity leads to feelings of we-ness and personal responsibility, which enhances self-restraint” (p. 20). De Cremer and van Vugt (Reference De Cremer and van Vugt1999) proposed that identifying strongly with the group increases cooperation in social dilemmas because people place extra high value on the group's collective project and welfare. They found that increasing group identification improved cooperation mainly among the members who started out oriented toward self and personal gain. Thus, personally endorsing the group's goals and welfare improved cooperation.
2.3.2. Aggression and mob violence
Riots, football hooliganism, violent protest demonstrations, gang battles, and similar phenomena epitomize some of the worst, most vicious and destructive tendencies of groups. In general, these are characterized by reducing individuality and submerging the self within the group. Le Bon (Reference Le Bon1896/1960) argued early on that the “group mind” was predisposed to simplistic thinking and violent action. Notions of the group mind led to a flurry of research on deindividuation, defined as a temporary reduction in self-awareness, personal responsibility, and evaluation apprehension, usually brought about by immersing the self in a group. Assorted findings linked the deindividuated state to aggressive, antisocial behavior (e.g., Beaman et al. Reference Beaman, Klentz, Diener and Svanum1979; Diener et al. Reference Diener, Fraser, Beaman and Kelem1976; Mann et al. Reference Mann, Newton and Innes1982; Nadler et al. Reference Nadler, Goldberg and Jaffe1982; Rogers & Ketchen Reference Rogers and Ketchen1979; Zimbardo Reference Zimbardo1969). Submersion in the group and loss of differentiated identity has been linked to lynch mob violence and wartime atrocities (Mullen Reference Mullen1986; Watson Reference Watson1973).
A meta-analysis by Postmes and Spears (Reference Postmes and Spears1998) concluded that deindividuation was mainly a matter of submerging oneself in the group and thus, following situational norms, such that when the group engages in bad behavior like cheating or stealing, deindividuation increases those tendencies. The primary effect of deindividuation was to reduce accountability, especially in enabling people to take illicit selfish benefits (e.g., cheating, stealing). They also found that problem behaviors increased with group size. All of those points are consistent with our analysis, including their conclusion that deindividuation effects are less a matter of inner states and more a matter of group or mob rule. Deindividuation thus submerges the self in the group, and one may go along with doing harmful, destructive things.
Converging evidence about the aggressive tendencies of group processes can be found in research on the interindividual intergroup discontinuity effect, as reviewed by Wildschut et al. (Reference Wildschut, Pinter, Vevea, Insko and Schopler2003). In laboratory studies with prisoner's dilemma and similar games, groups generally are less cooperative than individuals, in the sense that groups will choose more exploitative moves and fewer cooperative ones than individuals. Behavior becomes more antisocial and less cooperative when people are not being held individually responsible for their actions. In a group setting, individuals can support selfish and aggressive group decisions without taking responsibility, and if challenged they can say that their own support for such actions was simply a reaction to others' initiative. When people are identified, the nastiness of groups (relative to individuals) is mitigated (e.g., Schopler et al. Reference Schopler, Insko, Drigotas and Wieselquist1995). Likewise, simulated, anonymous jurors tended to make guilty judgments and recommend harsh punishments, but individually identified jurors were more lenient (Hazelwood & Brigham Reference Hazelwood and Brigham1998).
Mob violence, antisocial behavior, and the aggressive tendencies of groups (more so than individuals) are in large part a result of the submerging of the self into the group. A lack of personal responsibility and awareness of ethical standards – hallmarks of the deindividuated state – emerge when groups do not hold individuals accountable. Moral control is far improved when individuals come to the fore, in support of the second step of our model. In further support, one study involved a group context in which some group members believed they might have to be personally accountable for their actions, whereas others were not given accountability information. Aggression toward helpless victims was reduced in the accountability condition (Prentice-Dunn & Rogers Reference Prentice-Dunn and Rogers1982). Differential identification of individual selves – the literal opposite of deindividuation – is what enables group moral control.
2.4. Evidence for two complementary steps
We began by proposing that many of the most successful groups make use of two steps. The first involves building a sense of shared social identity, thus emphasizing sameness and cohesion among members. The second step involves increasing differentiation of roles and individuality.
A survey of managers at a Dutch bank about their middle-management teams provides evidence for both steps in our theory (Janssen & Huang Reference Janssen and Huang2008). A strong sense of shared identity promoted good citizenship behavior, such as helping and caring about others, but was irrelevant to creative performance. In contrast, a strong belief in one's distinctiveness (e.g., highlighting one's unique skills) was linked to high creativity but was irrelevant to citizenship. Thus, the first step of shared social identity promotes cohesion and helping, but the second step of differentiation contributes to group performance.
Spencer-Rodgers et al. (Reference Spencer-Rodgers, Hamilton and Sherman2007) studied perceptions of various groups. Being stable, having well-defined boundaries, and having highly similar members were characteristics ascribed to groups based on social categories (e.g., Californians, Jews, elderly). In contrast, task groups (e.g., juries, committees, theater troupes) were seen as much more differentiated, as well as more agentic and entitative. Thus, social perceptions affirm the importance of both steps. In particular, groups that have a job to do are seen as having higher levels of role differentiation, consistent with the view that differentiation facilitates performance.
Assorted evidence supports the value of shared group identity for promoting good citizenship, helping, harmony, and loyalty to the group (Kirkman & Shapiro Reference Kirkman and Shapiro2001; Moorman & Blakely Reference Moorman and Blakely1995; Penner et al. Reference Penner, Dovidio, Piliavin and Schroeder2005; Van Vugt & Hart Reference Van Vugt and Hart2004; Zdaniuk & Levine Reference Zdaniuk and Levine2001). Various findings have also shown that identifying with the social group increases contributions in public goods and sharing resources situations (e.g., De Cremer et al. Reference De Cremer, Van Knippenberg, Van Dijk and Van Leeuwen2008; Tyler & Degoey Reference Tyler and Degoey1995). The first step of building shared identity is undeniably useful for the group. Nonetheless, the second step of differentiation provides substantial advantages over the long run.
One possible proxy for the first step would be group cohesion, which seemingly expresses the members' embrace of the common group identity. A meta-analysis by Mullen and Copper (Reference Mullen and Copper1994) noted that there has been considerable debate about whether cohesion is linked to group performance at all. They concluded that the link is real but small. Moreover, the causal arrow points both ways, and the increase in cohesion following good performance is stronger than the (nonetheless still real) causal effect of cohesion on performance. The effect is also stronger in small groups than large ones. All of these findings are congenial to our analysis, which emphasizes that shared identity can occasionally help performance but is not a major factor, so that the second step (differentiation) is more important. We also suggest that competition among groups led to forming ever-larger groups, so although shared identity might have been sufficient with small groups, differentiation would become more important over time, as groups become larger. A small group, such as a team relay, may succeed by motivating members with shared identity even if there is no differentiation of roles, but the competitiveness of large organizations depends heavily on an effective system of differentiated roles and individual accountability.
Commons dilemma and other social dilemma patterns provide valuable evidence that both steps of group formation are important. They require a person to choose between immediately selfish, antisocial actions and enlightened self-interest through prosocial cooperation. Restraint and cooperation benefit the self only if others act the same, however, and so trust in the group is required. Findings show that identifying strongly with the group and embracing a shared social identity are helpful (e.g., Step 1 of our theory) – as are individual identification and the associated effects of responsibility and accountability (Step 2). These are not contradictory findings but rather complementary phenomena. The first step of group formation is embracing the shared social identity, which helps promote trust and willingness to cooperate. The second step is differentiation of selfhood, which enforces responsibility and motivates people to sustain the prosocial behavior that enables the entire group to benefit in the long run.
Abrams et al. (Reference Abrams, Wetherell, Cochrane, Hogg and Turner1990) reported a conformity study that manipulated both steps. They used an Asch conformity measure, in which confederates gave erroneous answers to a judgment task, and the measure was how much the true participants went along with those erroneous answers. The confederates were presented as belonging either to the participant's in-group or to an out-group, and the participant's responses were either public or private and anonymous. In private, the group made no difference, but conformity was high when participants made public responses in front of the in-group (and not in public responses to the out-group). Thus, shared identity led to poor performance by increasing conformity, presumably motivated by desire for acceptance based on similarity. Anonymity allowed people to think for themselves, thereby creating the benefits of differentiation.
Leaders can either suppress different perspectives by telling everyone what to do and think or solicit inputs from all and strive to integrate them. Lorinkova et al. (Reference Lorinkova, Pearsall and Sims2013) compared these styles in a laboratory study. Groups with directive leaders came together faster and performed best in the early rounds, whereas the groups with leaders who heeded different inputs floundered. After the fifth round, however, the performance results shifted heavily in favor of groups with leaders who sought to include all viewpoints. Thus, sameness based on cohesion as directed by a take-charge leader worked best at first, but in the long run, capitalizing on differentiation produced the best results.
A meta-analysis of the effects of work group diversity on innovation by Hülsheger et al. (Reference Hülsheger, Anderson and Salgado2009) reported separate analyses for background diversity (gender, ethnicity, age) and job-related diversity (differences in specialized function, skills, training, expertise, etc.). Background diversity is relevant to the first step because it complicates the formation of shared identity (Mannix & Neale Reference Mannix and Neale2005). Sure enough, this form of diversity failed to improve innovation and had, if anything, a negative effect. This supports the view that the first step benefits from common identity (which contributes only weakly, if at all, to performance). In contrast, diversity of skills and roles had a positive effect on innovation, producing better results for both the individual members and for the group as a whole.
We have reported multiple findings indicating that enthusiastic identification with the group can overcome individuals' selfish tendencies, thus strengthening the group (e.g., with social loafing). This, too, seems congenial to the argument that accountability becomes useful over time. Newly formed groups may often generate enthusiasm for the shared identity, so that all pitch in and work hard, but at some point, some members may be tempted to pursue a selfish agenda, and so accountability is needed. Consistent with that view, Van Vugt and de Cremer (Reference Van Vugt and De Cremer1999) found that instrumental leaders who punished noncontributing members had more effective groups than leaders who focused on simply building harmony in the group, particularly when group identification was low. When members identified strongly with the group, the two types of leaders were equally effective. Apparently, then, moral control of individuals is conducive to long-term success.
3. Discussion
We began by noting the paradoxical contradiction between two traditions of research on groups: Groups have been shown to be both better and worse than sets of individuals working alone. Much of the difference can be explained on the basis of differentiation of selves. A broad and diverse set of evidence converged to indicate that groups function better when members have differentiated identities than when individuality is lost as people blend into the group.
We suggested that groups form in two heuristic steps. The more fundamental one involves the construction of a shared group identity, which when embraced by individuals motivates them to work on behalf of the group. The second step (our main focus) involves a vast increase in performance and efficiency. Its key is not sameness but difference, insofar as different members use different skills to perform different roles in an interlocking, interactive system.
Differentiation does not contradict but rather builds on the sense of shared identity, which continues to be helpful. Indeed, we reviewed multiple lines of evidence that strong personal identification with the group (strong social identity) can motivate high effort and good behavior – very much unlike loss of individuality into the group, which had largely negative effects. Shared social identity is beneficial, whereas sameness in thought and action was often less helpful for the group than differentiation. Put another way, differences among group members are often crucial to the group's success. Groups may flourish by recognizing and capitalizing on those differences. In a highly competitive environment, they may need to do so to survive. Even some findings that emphasize identification with the group as beneficial also show the importance of individual identification, such as in procedural justice and accountability. Also, the historical and worldwide shift toward ever larger groups suggests that shared group identification will become less important (partly because big groups do not inspire such strong effects) whereas differentiation (e.g., specialization) will become increasingly important.
Again and again, we found that people contributed better as individually identified members and did worse when individual identity was downplayed or lost. In performance settings, people worked harder and did better insofar as they were individually identified, accountable, individually competing or otherwise evaluated, eligible for rewards contingent on individual performance and the like. Social loafing occurred when people felt like indistinguishable members of the group, especially in the sense that their own efforts and contributions would not be known to other group members. Knowing one's work would be individually identified to the other group members was a powerful cure for social loafing and other detrimental processes.
Another antidote to social loafing was a feeling of being indispensable: People did well even under relative anonymity if they believed that their individual contribution to the group was unique and necessary for the group's success. That signifies differentiation. If others could substitute for oneself with no penalty to the self, then one loafed. In social dilemma situations, groups managed their resources best when people were individually identified, whereas anonymous and nonaccountable systems tended to deplete resources and do poorly. When judgments had to be made, accountable members put in more thought and effort than others, and they also produced more output. Generally, keeping track of individual selves improved group performance in multiple ways. These furnish a basis for arguing that human selves evolved to facilitate successful performance by groups.
Turning to the informational functions of groups, we found evidence that groups benefit when members participate as separate, autonomous individuals. Pressure to conform to the group's consensus often yielded detrimental results, whereas independent thinking and even overt dissent often helped the group reach more accurate judgments and make better choices. The superiority of secret ballots over other systems of group decisions is one familiar sign of this phenomenon: The shield of anonymity frees the individual from having to conform to the group's (or the leader's) preferred views, thereby enabling each person to think and choose autonomously. Other work has shown that anonymity and independent thought enable groups to be wiser even than experts. Conversely, pressures to conform to the group can bias judgments (especially toward the group's favored views), can curtail information sharing, and may foster groupthink and its costly errors.
Moral behavior was also relevant. Morality generally encourages people to overcome selfish impulses and do what is best for the broader group (though this fact becomes complicated when groups engage in immoral activities). Higher moral principles and virtuous actions were generally facilitated by individual identification and accountability. In such cases, anonymity enabled people to indulge their prejudices, overconsume precious resources, and claim a share of collectively available benefits while contributing little or nothing to meeting the costs.
Indeed, the benefits of individuation go beyond what we have reviewed. People are more helpful when individually identified than when submerged in the group, as in research on diffusion of responsibility (Darley & Latané Reference Darley and Latané1968; Latané & Nida Reference Latané and Nida1981). Conversely, they are more aggressive when submerged in the group, as in cases of mob violence, football hooliganism, wartime atrocities, and tendencies for groups to be more destructive and antagonistic than identified individuals are. Individual identification of group members reduces these antisocial behaviors.
What matters is thus the relationship of the individual to the group, not the mere fact of anonymity or structure of the individual self. When individuals function as autonomous individuals who contribute to the group and are responsible to it, groups benefit. Systems bring gains but only if members play their distinctive, complementary roles. Individual selfishness is often an obstacle to effective group functioning, so the group either finds ways to restrain selfishness (e.g., with moral punishment) or to harness selfishness to the group goals. Indeed, the tortuous history of deindividuation research led to the conclusion that it is not an individual state of mind but a group phenomenon, involving submerging individual identity into the group (Postmes & Spears Reference Postmes and Spears1998). It is not the self acting on its own to exploit the group but rather the self participating in the group as a differentiated, yet cooperative member that yields the best results.
Many of these findings reflect the individual's desire for social approval and acceptance, and hence the group's ability to exert moral control over individual members by putting pressure for proper behavior. Publicly identified persons work hard in the expectation of being favorably evaluated by group members. Unfortunately, however, those same desires and pressures can undermine independent thought and therefore degrade the quality of group information processing, yielding poor judgments, bias, and bad decisions. As we proposed in the introduction, information processing is best served by having each individual think and conclude as an autonomous, independent self and then contribute as such to the group discussion. Even arguing different sides of an issue is often valuable. We cited evidence that groups benefit from dissent, even in cases in which no member initially supports the best decision – because arguing helps to air all relevant facts, so that the group can come around to the best answer.
The group uses individually identified, differentiated selves as a tool for controlling behavior. The group works best if it makes many individual members do what they are supposed to do. It accomplishes that in part with rewards and punishments, but those depend on accountability and selfhood. With appropriate rewards and punishments, the group can increase effort and improve the moral quality of behavior. But it can also suppress independent thought, thereby degrading the informational quality of the group's knowledge base and decision processes.
To be sure, not all manifestations of differentiation are beneficial. Narcissism, in particular, may produce ill effects insofar as people overvalue themselves and feel entitled to exploit others (e.g., Twenge & Campbell Reference Twenge and Campbell2003). There also are cases in which anxiety over evaluation can inhibit participation in groups and reduce overall performance (e.g., Camacho & Paulus Reference Camacho and Paulus1995). And excessive diversity in groups, especially diversity of ethnicity or background, can hamper communication, reduce cohesiveness, and otherwise impair performance (Mannix & Neale Reference Mannix and Neale2005). It is possible to regard such instances as too much of a good thing or as irrelevant to the basic point that groups mostly benefit from differentiated selves. Because of space constraints, we sought to make the case for the view that differentiation of identity is useful, rather than to survey all findings, and so we have not dealt with every possible counterexample. We think that even if differentiation is not invariably helpful to groups, it is helpful far more often than not, which is sufficient for our argument that one basic function of the human self is to facilitate group processes.
3.1. Implications for self theory
The view that groups benefit from differentiated selves offers a possible basis for a theory about the functional origins of human selfhood. If our view is correct, human selfhood emerged not out of some peculiar inner dynamic such as motivational or brain processes (though those presumably mediated the emergence of human selfhood), but as a vital adaptation to capitalize on the immense potential advantages of group life and group action. Indeed, some analyses have concluded that the very survival of the species depended on the development of advanced social systems (i.e., with division of labor and economic trade) based on differentiated selves (Horan et al. Reference Horan, Bulte and Shogren2005). With their large bodies and brains, individual Neanderthals would have competed effectively against individual humans – but Neanderthals were unable to match the human gift for developing social systems. Collectively they were unable to compete with modern humans' Cro-Magnon ancestors, and they lost out and became extinct.
One perennial puzzle in self theory is why human selfhood is so much more advanced and complex than what has been observed in any other species. Our findings suggest that a major part of the answer lies in the usefulness of differentiated selves for human groups – especially large ones. Larger groups permit more complex and thus more differentiated systems than small ones, so role identities can be more specialized. (Hence, many selves are labeled with names that refer to occupational roles; e.g., Shoemaker, Smith, Baumeister, Tailor). Even shared aspects of identity may gain complexity as groups expand. As Moffett (Reference Moffett2013) explained, humans and a few insects are the only species that have cooperative groups larger than about 150 members with strict boundaries. (Large grazing herds have casual boundaries, such that animals can move from one herd to another without much ado.) The insects accomplish this without highly differentiated selves: An ant can apparently not recognize a particular other ant, though it can distinguish between an ant from its own versus a rival colony. Humans, however, build their large groups with differentiated individual identities, thus permitting much more complex systems to emerge.
We assume that groups using complex social systems had competitive advantages over groups lacking such systems. Systems are made up of roles, and so it was adaptive for human selves to become able to perform these roles. Insofar as the human self evolved to facilitate cultural groups, it had to acquire the capability to operate in such systems. In other words, human selfhood has to furnish players for the differentiated roles that populate such systems.
Recent efforts to understand the essential nature of human selfhood have struggled to locate it, despite mountains of data about various concepts and processes of the self. The lack of any specific brain seat for the self has led some to speculate that the self is an illusion or fiction (Metzinger Reference Metzinger2009), a view echoed on other conceptual grounds by some social psychologists (Swann & Buhrmester Reference Swann and Buhrmester2012), who define it as a functional fiction. Self-concepts do indeed often contain liberal doses of fiction, but the flexible capacity to perform real roles in complex real groups may be a vital basis for genuine selfhood.
Thus, our review offers another way to ground self theory. Complex social systems depend on differentiated identities and in fact benefit most from a high level of differentiation. The human brain may not be organized with a central, controlling “self” in it, but it learns to operate a self within the social system. The present evidence indicates that human groups derive advantages from having differentiated selves. One may therefore speculate that human minds evolved the capacity to capitalize on those advantages. Individuals would have benefited by joining groups composed of members with differentiated selves because these would likely have outperformed less differentiated groups. Survival and reproduction could thus have benefited from developing the capacity to participate in groups with differentiated selves. In this view, the self is not fiction – it is a reality, albeit a social one, that an individual physical body learns, acquires, and becomes.
Our findings also suggest which aspects of selfhood are most conducive to effective group functioning. Agentic control of effort, autonomous thought and judgment, and moral responsibility were all repeatedly found to benefit group outcomes. If the human self did partly evolve to facilitate human group processes, those three aspects would likely have been central.
3.2. Implications for group theory
Our analysis offers one resolution to the seeming contradiction in the literature on groups. Two distinguished traditions of empirical research have documented at length how groups are sometimes much more and better than the sums of their individual members – but are at other times much less and worse. Differentiated selfhood provides one vital conceptual key to account for what enables the positive outcomes, and its absence (submerging individuality in the group) helps explain many negative ones.
We reiterate that differentiated selfhood is not the opposite of identification with the group. Shared group identity promotes cohesion and various prosocial behaviors. The benefits of role differentiation may often combine with the enthusiastic embrace of shared identity for best results, as in a sports team with strong team spirit plus highly differentiated task roles. Identifying with the group can in principle be based on a highly personal, individual decision or could be a matter of losing identity into the group (e.g., Swann et al. 2012). Our findings also broadly fit the heuristic division of the group formation process into two steps, in which the first builds a shared identity and the second differentiates roles. Shared identity may be quite helpful, especially at first, but in the long run and perhaps in larger, more impersonal groups, differentiation becomes vital for effective group functioning.
4. Conclusion
Sometimes groups are much more than the sum of their parts, sometimes much less. Individually identified separate selfhood is one key difference. Most of the bad effects of groups (e.g., social loafing, collective resource depletion) come when the individual self is lost or forgotten as identity is submerged in the group. When group members blend together, responsibility is lost, enabling extreme and antisocial behaviors. Then the group is less or worse than the sum of its individuals. In contrast, when members are accountable and responsible, and they fill different roles in interacting systems (family, the local economy, division of labor and specialized expertise), then the system gain can make the group more than the sum of its parts.
The very definition of group invokes some sameness: The members all belong to the same group and presumably share some goals, values, and identity. In practice, moreover, many groups push for sameness on many dimensions. In order to thrive, however, groups may need to go beyond the sameness of their members and capitalize on differences. Shared social identity is useful, but lack of individual identification can be costly. Differentiated selves and accountable individuality provide keys to the immense success of human groups.
Target article
Are groups more or less than the sum of their members? The moderating role of individual identification
Related commentaries (29)
Beyond old dichotomies: Individual differentiation can occur through group commitment, not despite it
But is it social? How to tell when groups are more than the sum of their members
Considering the role of ecology on individual differentiation
Differentiated selves can surely be good for the group, but let's get clear about why
Differentiated selves help only when identification is strong and tasks are complex
Differentiation of selves: Differentiating a fuzzy concept
Disputing deindividuation: Why negative group behaviours derive from group norms, not group immersion
Group and individual as complementary conceptual categories
Group behavior in the military may provide a unique case
Group effort in resuscitation teams
Group members differ in relative prototypicality: Effects on the individual and the group
Group membership: Who gets to decide?
Groups need selves, but which selves? Dual selves in groups and the downsides of individuation
How group members contribute to group performance: Evidence from agent-based simulations
Humans are not the Borg: Personal and social selves function as components in a unified self-system
Identity matters to individuals: Group assessment cannot be reduced to collective performance
Member differentiation and group tasks: More than meets the eye
Not even wrong: Imprecision perpetuates the illusion of understanding at the cost of actual understanding
Reputational concerns as a general determinant of group functioning
Roles and ranks: The importance of hierarchy for group functioning
Social identification is generally a prerequisite for group success and does not preclude intragroup differentiation
Social, not individual, identification is the key to understanding group phenomena
Solved paradoxes and old hats? The research needed on differentiated selves
Task specificity and the impact on both the individual and group during the formation of groups
The hows and whys of “we” (and “I”) in groups
The subtle effects of incentives and competition on group performance
The unique role of the agent within the romantic group
Vicarious contagion decreases differentiation – and comes with costs
We agree and we disagree, which is exactly what most people do most of the time
Author response
Differentiating selves facilitates group outcomes