Limits of Group Learning

The general pattern found on a variety of tasks is one of groups being superior to individuals (certainly the “average” individual, and sometimes even the “best” indi- vidual—see Reagan-Cirincione, 1994; Weldon & Bellinger, 1997) and of groups improving with experience. Groups are better than individuals on recall (Hinsz, 1990) and induction tasks (Laughlin & Hollingshead, 1995). Groups become better at sharing information (Wittenbaum, 1996) and developing more complex under- standings of phenomena (Gruenfeld & Hollingshead, 1993) as they gain experience. With experience, groups also become better at identifying expertise and who knows what (Liang et al., 1995) and at recognizing and accepting the correct solution pro- posed by a group member (Laughlin & Hollingshead, 1995). Group performance improves as groups gain experience (Argote, Insko, Yovetich, & Romero, 1995; Goodman & Leyden, 1991; Guetzkow & Simon, 1955; Shure, Rogers, Larsen, & Tassone, 1962).

Although the general pattern is one of group superiority to individuals, there are important examples where groups make very poor decisions. Janis (1972) vividly described examples of fundamentally flawed group decisions with disastrous conse- quences and attributed the flaws to the groupthink phenomenon. Controversy has swirled around the groupthink phenomenon, the conditions under which it occurs and whether it leads to defective outcomes (see Aldag & Fuller, 1993, for a review). Several of the conditions hypothesized to lead to groupthink arguably prevent groups from learning. For example, a leader who actively promotes his or her opin- ions might prevent others from speaking up or voicing their views. Similarly, a lack of norms about the importance of incorporating all members’ inputs in decisions could discourage members from sharing information they possess. Thus, group- think may block the learning processes that generally occur in groups as members gain experience working together. In a related vein, Argyris (1992) demonstrated how the defensive routines organizations invoke to avoid embarrassment or threat can prevent organizational learning (see also Argyris & Schon, 1978).

Janis (1972) described how the same group of policy makers who made very poor decisions in the Bay of Pigs fiasco functioned much more effectively in the Cuban Missile Crisis after they had gained experience working together. By con- trast, Kim (1997) found that groups with both experience on a similar task (task experience) and experience working together (team experience) performed more poorly than their counterparts lacking both types of experience in a laboratory study. The poor performance of groups with both task and team experience was due in part to their tendency to focus on information that members held in common rather than discuss unique information that different members possessed.

Katz (1982) found a non-monotonic inverted U-shaped relationship between group “longevity” (the average amount of time group members had worked together) and the performance of Research and Development (R&D) groups. Performance initially increased and then decreased with increases in the average group experi- ence. Initially, performance improved as group members learned how to communicate and coordinate their activities. The decrease observed in performance at high levels of experience was attributed to the groups becoming too inwardly focused and not interacting with or learning from external sources. These results underscore both the benefits and the costs of experience. Increases in experience can enhance group performance significantly by providing group members with opportunities to learn how to work together effectively. If increases with experience are also associated, however, with becoming isolated from external sources of knowledge, performance decrements can occur.

Although there is considerable evidence of individual performance improving with experience, there is also evidence that individuals have difficulties drawing appropriate inferences from their experience (Dawes, 1988; Einhorn & Hogarth, 1986; Tversky & Kahneman, 1974). Not only does the performance of most groups improve with experience, groups seem to improve in interpreting their experience (e.g., see Gruenfeld & Hollingshead, 1993).

How can these conflicting viewpoints of the role of experience in individual and group performance be reconciled? A possible reconciliation would be to demon- strate that groups are better than individuals at learning from experience. Comparing how individuals and groups learn from experience would extend previous compari- sons of individual and group performance (Hill, 1982; Maier, 1967) in interesting ways. In their analysis of group versus individual accuracy, Gigone and Hastie (1997) concluded that “groups tend to outperform individuals when the judgment policies of their most accurate members are both more accurate than the judgment policy of the statisticized groups and discernible by other members” (p. 163). As noted earlier in this chapter, perceptions of the expertise of other group members have been found to become more accurate as groups gain experience. This may explain an advantage groups have relative to individuals at learning from experience.

Research has been done comparing individual and group biases that is also some- what relevant for this issue (e.g., see Argote, Seabright, & Dyer, 1986; Kerr, MacCoun, & Kramer, 1996; Tindale, 1993). This work examines whether groups amplify or cancel decision-making biases observed at the individual level (see Kahneman, Slovic, & Tversky, 1982, for a discussion of biases at the individual level). Most of this work has been conducted at one point in time. Hence it does not directly address questions of learning but rather provides comparisons of group versus individual judgment.

Research comparing group and individual biases has yielded contradictory findings. Some studies have shown that groups eliminate biases observed at the individual level, whereas others have reported that groups amplify individual biases. Whether groups amplify or cancel biases depends on the nature of the bias. If the bias is not systematic, group discussion will serve to weaken the bias because members will be exposed to other points of view that are less biased or biased in a different direction (see also Gigone & Hastie, 1997). Conversely, if the bias is sys- tematic and widely shared, group discussion is likely to amplify the bias by expos- ing members to others who hold the same bias and thereby, increasing confidence in its correctness. Thus, groups may be better than individuals at learning if group members have diverse views that are utilized effectively. Even when individual biases are systematic, groups may be able to outperform their best member if their discussion is facilitated to ensure that all group members participate and if members are provided feedback as a group (Reagan-Cirincione, 1994).

The context in which learning occurs within groups also helps to explain varia- tion in the extent to which groups learn from experience. A culture of “psychologi- cal safety,” where group members feel free to express their ideas, has been found to facilitate learning (Edmondson, 1999). A “learning” orientation, in which members emphasize learning, has been found to lead to better performance than a “perform- ing” orientation, in which members compare their performance to that of other units (Bunderson & Sutcliffe, 2003). Team structures characterized by specialization, formalization, and hierarchy have been found to facilitate team learning because those bureaucratic structures increased information sharing and reduced conflict (Bunderson & Boumgarden, 2010). Research on how the context interacts with experience to affect learning processes and outcomes is very promising.

More research is needed on the question of how groups (and organizations) learn from experience and the conditions under which their performance improves with experience. Determining the answer to this question will require longitudinal stud- ies that examine groups over time. It will also require studying many groups that differ in interesting ways. This research strategy will enable us to specify more fully the conditions under which group learning occurs. Characteristics of the context in which the group is embedded are likely to condition the extent to which group learning occurs.

Source: Argote Linda (2013), Organizational Learning: Creating, Retaining and Transferring Knowledge, Springer; 2nd ed. 2013 edition.

Leave a Reply

Your email address will not be published. Required fields are marked *