SUMMARIZE THE ARTICLE BELOW

:

Teams of people working together for a common purpose have been a centerpiece of human social organization ever since our ancient ancestors first banded together to hunt game, raise families, and defend their communities. Human history is largely a story of people working together in groups to explore, achieve, and conquer. Yet, the modern concept of work in large organizations that developed in the late 19th and early 20th centuries is largely a tale of work as a collection of individual jobs.

A variety of global forces unfolding over the last two decades, however, has pushed organizations worldwide to restructure work around teams, to enable more rapid, flexible, and adaptive responses to the unexpected. This shift in the structure of work has made team effectiveness a salient organizational concern. Teams touch our lives everyday and their effectiveness is important to well-being across a wide range of societal functions. There is over 50 years of psychological researchliterally thousands of studiesfocused on understanding and influencing the processes that underlie team effectiveness. Our goal in this monograph is to sift through this voluminous literature to identify what we know, what we think we know, and what we need to know to improve the effectiveness of work groups and teams. We begin by defining team effectiveness and establishing the conceptual underpinnings of our approach to understanding it. We then turn to our review, which concentrates primarily on topics that have well-developed theoretical and empirical foundations, to ensure that our conclusions and recommendations are on firm footing. Our review begins by focusing on cognitive, motivational/affective, and behavioral team processesprocesses that enable team members to combine their resources to resolve task demands and, in so doing, be effective. We then turn our attention to identifying interventions, or levers, that can shape or align team processes and thereby provide tools and applications that can improve team effectiveness. Topic-specific conclusions and recommendations are given throughout the review. There is a solid foundation for concluding that there is an emerging science of team effectiveness and that findings from this research foundation provide several means to improve team effectiveness. In the concluding section, we summarize our primary findings to highlight specific research, application, and policy recommendations for enhancing the effectiveness of work groups and teams. INTRODUCTION Houston, we've had a problem. Apollo 13 was more than halfway on her journey to Earth's moon on what was to have been a routine mission to collect samples when, suddenly, the mission and the lives of the crew were in grave jeopardy. One of the spacecraft's two oxygen tanks exploded, blowing out the entire side of the service module and damaging the remaining oxygen tank. Within 3 hours, all oxygen stores were depleted, and the craft lost water, electrical power, and propulsion. The situation was critical, time was short, and there was no margin for error. A team of NASA engineers was hastily assembled. Their mission: problem-solve, adapt, and invent a way for the crew to survive and to pilot their damaged spacecraft back to Earth. The team was successful, transforming a potentially disastrous mission into a legend of effective teamwork (NASA Goddard Space Flight Center, n.d.). Jump to Nepal, deep in the heart of the Himalaya Mountains. Several international teams were mounting that annual campaign of human striving and accomplishment, attempting to reach the summit of Mt. Everestan intrinsically irrational act (Krakauer, 1997, p. xvii). The teams were led by renowned mountaineers, but this season on Everest turned out to be the most disastrous one of all time. On one team, of five teammates who reached the peak, four, including the veteran leader, died. Nine climbers from four other expeditions also perished. Before the month was out, 16 climbers lost their lives attempting to reach the treacherous summit. Although the harsh, unforgiving, and constantly changing environment played a major role in this tragedy, the perilous conditions were exacerbated by failures of team leadership, coordination, and communication (Krakauer, 1997). Teams of people working together for a common cause touch all our lives. From everyday activities like air travel, fire fighting, and running the United Way drive to amazing feats of human accomplishment like climbing Mt. Everest and reaching for the stars, teams are at the center of how work gets done in modern life. Although how teams function is often beneath the level of everyday awareness, unexpected successes, such as Team USA's winning of the Olympic Gold Medal for Hockey, and failures, such as FEMA's sluggish response to hurricane Katrina, make team functioning and team effectiveness highly salient. Failures of team leadership, coordination, and communication are well-documented causes of the majority of air crashes, medical errors, and industrial disasters. They have also been implicated in many political and military catastrophes, including the miscalculated Bay of Pigs invasion, the mistaken downing of a civilian airliner by the USS Vincennes, the failure of the USS Stark to take defensive action against a hostile missile attack, and the failure to prevent the tragedy of 9/11. Our point is simple: teams are central and vital to everything we do in modern life. Our purpose in this monograph is to elucidate what more than 50 years of research on small groups and teams can tell us about the processes that contribute to team effectiveness and, based on that knowledge, to identify leverage points that can be used to make teams more effective. Review Focus and Structure Organizations around the world are well along a decade-and-a-half evolution in the design of workshifting from individual jobs in functionalized structures to teams embedded in more complex workflow systems (Devine, Clayton, Phillips, Dunford, & Melner, 1999; Lawler, Mohrman, & Ledford, 1992, 1995; Mathieu, Marks, & Zaccaro, 2001). A variety of forces are driving this shift. Increasing competition, consolidation, and innovation create pressures for skill diversity, high levels of expertise, rapid response, and adaptability. Teams enable these characteristics (Kozlowski, Gully, Nason, & Smith, 1999). The increasing penetration of computers into all facets of the workplace coupled with broadband communication allows teams to be located together or distributed across time and space (Bell & Kozlowski, 2002b). Multicultural teams linked across the globe by technology are on the rise. Concomitant with this shift in the organization of work is the shift in research focus from the study of small interpersonal groups in social psychology to the study of work teams in organizational psychology. This shift in the core of team research was explicitly recognized by Moreland, Hogg, and Hains (1994), who noted the relative decline of group research in social psychology, and by Levine and Moreland (1990), who concluded that small-group research is alive and well and living elsewhere [outside the confines of social-psychology laboratories] (p. 620). At least seven major reviews of the work-team literature in organizational psychology appeared between 1990 and 2000 (see Bettenhausen, 1991; Cohen & Bailey, 1997; Gully, 2000; Guzzo & Dickson, 1996; Guzzo & Shea, 1992; Hackman, 1992; Sundstrom, McIntyre, Halfhill, & Richards, 2000).1 More recent reviews of work-team research (Ilgen, Hollenbeck, Johnson, & Jundt, 2005; Kozlowski & Bell, 2003) reflect the emerging perspective of work teams as dynamic, emergent, and adaptive entities embedded in a multilevel (individual, team, organization) system (cf. Arrow, McGrath, & Berdahl, 2000; Kozlowski et al., 1999; Marks, Mathieu, & Zaccaro, 2001). That is, teams are complex dynamic systems that exist in a context, develop as members interact over time, and evolve and adapt as situational demands unfold. Dynamic complexity; emergent team processes and phenomena; and development, evolution, and adaptation are key themes that will be reflected in our review. We will describe what we mean in more detail in the next section, but a brief overview here is useful because it sets the structure for our focus and approach. As illustrated in Figure 1, a team is embedded in a broader system context and task environment that drives team task demands; that is, the task requirements necessary to resolve the problem or situation presented by the environment and the load placed on team members' resources. A dynamic, shifting, and complex environment creates commensurate team task demands that members have to resolve though a coordinated process that combines their cognitive, motivational/affective, and behavioral resources. As Figure 1 shows, this process is cyclical and reciprocal. When team processes are aligned with environmentally driven task demands, the team is effective; when they are not, the team is not. Our approach is guided by this basic heuristic and the focal points of our review are captured in the highlighted portion of the figure. We first consider team effectiveness as a dynamic process. We next review the research base to identify critical team processes and emergent states that contribute to team effectiveness. Having established that research foundation, we then consider factors that can influence, shape, and make appropriate team processes. Thus, our basic questions in this review are: What are the key team processes and emergent states that influence team effectiveness? How can these processes and states be leveraged to better create, develop, and manage effective work teams? Fig. 1. Conceptual framework and review focus. The figure illustrates that environmental dynamics and complexity drive team task demands; team processes and emergent states align team-member resources to resolve task demands and yield team effectiveness; and team outputs (effectiveness) reciprocally influence the environment, in an ongoing cycle. The focus of this report is shaded: team processes and emergent states; and the factors that shape, leverage, or align them. There has been well over a half century of research in both social psychology and organizational psychology on small groups and teams and related topics. Virtually all of the research papers close with the obligatory acknowledgement that more research is needed. And, while it is true that there is much we psychologists do not yet know about team effectiveness, there is much that we do. There is a substantial knowledge base. The real challenge is sifting through this vast literature to isolate those promising team processes that reliably influence team effectiveness and that can also be shaped by deliberate intervention. Our intent in this review is to focus on those key areas in which theory and research findings are well developed and therefore provide a solid substantive basis for actionable recommendations. Although this review is ultimately guided by our own theoretical perspectives, empirical research, and professional judgment, we relied on three primary strategies to identify key areas. First, we sought research topics that were sufficiently mature that they had been the target of one or more meta-analytic reviews.2 Meta-analytic findings provide a quantitative foundation for our most forceful conclusions and recommendations. Second, we sought topics that, though not meta-analyzed, had been the subject of substantial systematic empirical research. For these areas, our conclusions and recommendations are strong but not unequivocal. Finally, we also considered areas that reflect emerging theory and promising, though not yet extensive, research support. The potential of these areas is more a matter of our judgment and the conclusions and recommendations are therefore intended to be more circumspect. As in any such endeavor, our decisions about what to include and exclude will not please everyone. We offer apologies in advance to all whose work may have been overlooked. We now begin by articulating our theoretical perspective, as it highlights important themes of our review and its structure. We next review research on team processes and emergent states, giving recommendations about those that are actionable and those that require further development and research attention. We then shift to an identification of several potent intervention "levers" that can shape team processes. If you want to know how to enhance team effectiveness, this is how it can be accomplished. The policy implications in these two core sections are self-evident. Finally, we close with a summary of our many recommendations and more general policy implications for the enhancement of team effectiveness. The Nature of Teams and Team Effectiveness What Is a Team? A team can be defined as (a) two or more individuals3 who (b) socially interact (face-to-face or, increasingly, virtually); (c) possess one or more common goals; (d) are brought together to perform organizationally relevant tasks; (e) exhibit interdependencies with respect to workflow, goals, and outcomes; (f) have different roles and responsibilities; and (g) are together embedded in an encompassing organizational system, with boundaries and linkages to the broader system context and task environment (Alderfer, 1977; Argote & McGrath, 1993; Hackman, 1992; Hollenbeck et al., 1995; Kozlowski & Bell, 2003; Kozlowski, Gully, McHugh, Salas, & Cannon-Bowers, 1996; Kozlowski et al., 1999; Salas, Dickinson, Converse, & Tannenbaum, 1992). What Is Team Effectiveness? The conceptualization of team effectiveness that has shaped the last 40 years of theory and research is based on the logic of an input-process-output (I-P-O) heuristic formulated by McGrath (1964; cf. Gladstein, 1984; Salas et al., 1992). In this framework, inputs refer to the composition of the team in terms of the constellation of individual characteristics and resources at multiple levels (individual, team, organization). Processes refer to activities that team members engage in, combining their resources to resolve (or fail to resolve) task demands. Processes thus mediate the translation of inputs to outcomes. Although team processes are by definition dynamic, they are most typically addressed in static termsas constructs that emerge over time (i.e., emergent states) as team members interact and the team develops (Kozlowski et al., 1999; Marks et al., 2001). Output has three facets: (a) performance judged by relevant others external to the team; (b) meeting of team-member needs; and (c) viability, or the willingness of members to remain in the team (Hackman, 1987). These tripartite facets capture the prevalent conceptualization of team effectiveness. Although McGrath's heuristic is a useful organizing frameworkit was developed to organize the research literature on small groups circa 1964it was not intended to be a theory or a formal causal model of team effectiveness. It has, nonetheless, been frequently interpreted as a model to be tested. We think that while the I-P-O model is a useful organizing heuristic, treating it as a causal model encourages taking a limited and static perspective on team effectiveness and the dynamic processes that underlie it. A Dynamic View of Team Processes and Effectiveness We adopt a more contemporary perspective that has evolved over the last decade, which conceptualizes the team as embedded in a multilevel system that has individual, team, and organizational-level aspects; which focuses centrally on task-relevant processes; which incorporates temporal dynamics encompassing episodic tasks and developmental progression; and which views team processes and effectiveness as emergent phenomena unfolding in a proximal task- or social context that teams in part enact while also being embedded in a larger organization system or environmental context (Arrow et al., 2000; Ilgen et al., 2005; Kozlowski & Bell, 2003; Kozlowski et al., 1999; Kozlowski, Gully, McHugh et al., 1996; Marks et al., 2001). We now briefly highlight the key themes of this extension, elaboration, and refinement of the I-P-O heuristic; these themes play an important role in our conceptualization and organization of the literature and in our effort to make actionable recommendations based on that literature. Multilevel System Context Individual team members comprise the team as a collective entity, an entity that also serves as the social context that influences individual members (Hackman, 1992). Moreover, as illustrated in Figure 1, team members and work teams are embedded in a broader organizational system and task environment that drives the difficulty, complexity, and tempo of the team task. The interactions are reciprocal in that team performance outputs resolve task demands emerging from the surrounding system or environment and change the state of the system or environment in some fashion. These changes can shift unexpectedly and the team must adapt to the changing demands. Thus, it is necessary to understand the system context and linkages across multiple levelsindividual, team, organizationas key sources of contingencies or demands on the team that necessitate aligned team processes (Kozlowski et al., 1999; Kozlowski, Gully, McHugh et al., 1996). The degree to which teams are embedded in or tightly linked to the organizational system or a dynamic task environment can vary. Some teams or small units, while part of an organizational system, are more tightly linked to a dynamic task environment that is their dominant embedding context for task activity. As an example, consider a surgical team in the operating room (OR) where what is happening with the patient right now (e.g., dropping blood pressure, respiratory difficulty, erratic heartbeat) defines the task environment, which then drives team task demands and team-member activity. Relative to the broader organizational-system context (e.g., new policies adopted by hospital administration), the task environment is the primary context in which the OR team is embedded. The situation is similar for aircrews or firefighting teams, in which the task environment (i.e., take-offs, storms, and landings, or fire, fuel, wind, and humidity, respectively) is the primary embedding context. For other teams, the broader organizational system is the primary context. A cross-functional project team making a recommendation to management on product development or a top-management team (TMT) revising organizational strategy to meet stiff competition are more tightly included in the organizational system as the primary embedding context. The Team Task The central focus on what teams have to dotheir taskis the key factor that distinguishes a social-psychological perspective on the study of teams, in which the task is merely a means to prompt interpersonal interaction, from an organizational perspective, in which the task is the source of goals, roles, and task-based exchanges. For the latter, interpersonal interaction is relevant, but it is in the background rather than the foreground. The team task determines two critical issues. First, it sets minimum requirements for the resource poolthe constellation of team-member individual differences and capabilitiesthat is available across team members. If members collectively lack necessary knowledge, skills, abilities, or resources to resolve the team task, the team cannot be effective. Second, the team task determines the primary focus of team-member activities. Our focus is on teams that primarily do things (e.g., action or production teams) and that, in the process of striving toward and accomplishing goals, also have to make decisions (e.g., project teams or TMTs) and create, invent, and adapt solutions to resolve task-driven problems.4 Thus, the team task determines the workflow structure and coordination demands (i.e., exchanges of behavior, information, etc.) necessary for accomplishing individual and team goals and resolving task requirements. Team processes as emergent constructs or "states" are a way to capture coordination of team-member effort and factors relevant to it, as well as the alignment of team processes with task demands. In that sense, appropriately aligned team action processes are critical enablers of team effectiveness (Kozlowski et al., 1999; Kozlowski, Gully, McHugh et al., 1996; Marks et al., 2001; Salas et al., 1992). Time Team processes develop and unfold over time (McGrath, 1991). The extent to which team processes align with task demands is a function of team learning, skill acquisition, and development. Key temporal dimensions include (a) task cycles or episodes that entrain the team to task dynamics by making specific, iterative, and repeated demands on team processes (Ancona & Chong, 1996; Kozlowski, Gully, McHugh et al., 1996; Marks et al. 2001) and (b) linear development of indeterminate duration across the team's life cycle of formation, development, maintenance, and decline/dissolution (Tuckman, 1965; Kozlowski et al., 1999). The important points here are that team tasks are not fixedthey vary in their demands on team processesand that team process capabilities are not fixedthey compile and improve as team members accrue experiences and learn how to work together better. And, although processes are clearly dynamic, over time stable process constructs, or what Marks et al. (2001) call emergent states, develop, providing a means to capture or summarize team processes. Our Approach We begin our review by establishing what we know about critical team processes and emergent states and how they contribute to team effectiveness. Our purpose is to identify processes that have well-established, research-based linkages to team effectiveness that therefore should be targets for interventions to improve team functioning. We then highlight research-based interventions that can leverage team processes. With that foundation, we provide recommendations for enhancing team processes and effectiveness and offer suggestions for future research and applications. TEAM PROCESSES, EMERGENT STATES, AND EFFECTIVENESS Conceptually, process captures how team members combine their individual resources, coordinating knowledge, skill, and effort to resolve task demands. Team effectiveness (i.e., performance evaluated by others, member satisfaction, viability) is an emergent result that unfolds across levels (individual to dyadic to team) and over time. This perspective on team processes is clearly dynamic, but it is also the case that the repeated interactions among individuals that constitute processes tend to regularize, such that shared structures and emergent states crystallize and then serve to guide subsequent process interactions. Process begets structure, which in turn guides process. Allport (1954) described this reciprocal nature of process and structure in terms of ongoings, Katz and Kahn (1966) in terms of role exchanges, Kozlowski and Klein (2000) in terms of "emergent phenomena," and Marks et al. (2001) in terms of emergent states. Thus it is important to appreciate that while processes are dynamic and therefore difficult to capture in real time, they yield collective cognitive structures, emergent states, and regular behavior patterns that have been enacted by, but also guide, team processes. In that sense, team cognitive structures, emergent states, and routinized behavior patterns are the echoes of repeated process interactions (Kozlowski & Klein, 2000) and, hence, are indicative of the nature and quality of dynamic team processes. Following the structure set by Kozlowski and Bell (2003) and Ilgen et al. (2005) in their previous reviews, we classify team processes and their echoes according to whether they are cognitive, affective/motivational, or behavioral in nature.5 Team Cognitive Processes and Structures Small-group research has a long tradition of studying cognitive constructs such as group norms and role expectations that guide interpersonal interactions among team members. While not denying the importance of interpersonal interactions, research in organizational psychology has tended to address cognitive constructs that are more focused on guiding task-relevant interactions among team members. Indeed, work groups and teams have been characterized as processors of information (Hinsz, Tindale, & Vollrath, 1997). We focus on a set of team cognitive constructs that represent the structure of collective perception, cognitive structure or knowledge organization, and knowledge or information acquisitionconstructs that have amassed a sufficient research foundation to support their value for enhancing team effectiveness. These collective constructs include unit and team climate, team mental models and transactive memory, and team learning. Unit and Team Climate The notion of climate as an interpretation of the group situation or environment can be traced back to early work by Lewin, Lippitt, and White (1939), with much research and development over the intervening decades (see Forehand & Gilmer, 1964; James & Jones, 1974; Ostroff, Kinicki, & Tamkins, 2003 for comprehensive reviews). Contemporary theory and research regard climate as cognitively based, descriptive, interpretive perceptions of salient features, events, and processes (James & Jones, 1974) that characterize the strategic imperatives (Schneider, Wheeler, & Cox, 1992) of the organizational and team context. Although such perceptions originate within the person, exposure to strong strategic imperatives or situations (Gonzlez-Rom, Peir, & Tordera, 2002); perceptual filtering and interpretation by leaders (Kozlowski & Doherty, 1989); and social interaction, sharing of perspectives, and collective sense making (Rentsch, 1990) can shape a convergent emergent process that yields consensual, collective climate perceptions within teams, larger units, or organizations (Kozlowski & Klein, 2000). Research has established the emergence of collective climates at the organizational level (Kozlowski & Hults, 1987), in smaller units such as bank branches (Schneider & Bowen, 1985), and within teams embedded in organizations (Anderson & West, 1998; Hofmann & Stetzer, 1996). There is a large and growing research base demonstrating that collective climate relates to the performance, member satisfaction, and viability facets of individual, team, and unit effectiveness. For example, Schneider and Bowen (1985) showed that a shared, collective climate in which service was the salient strategic imperative predicted customers' satisfaction with their bank branch. Kozlowski and Hults (1987) demonstrated that a shared organizational climate of the strategic imperative to stay technically up-to-date and innovative predicted individual performance, continuing education activities, and positive job attitudes for engineers in organizations pressured by technological competition and change. Anderson and West (1998) showed that a team climate for innovation predicted overall team innovativeness, novelty of innovations, and number of innovations. Hofmann and Stetzer (1996) showed that a team climate for safety predicted safety-related behaviors and actual accident rates in a chemical plant where high-reliability performance is necessarily a high priority. At the individual level, a recent path analysis of correlations estimated via meta-analysis demonstrated that climate perceptions influence job performance, well-being, and job withdrawal (e.g., intentions to quit; Carr, Schmidt, Ford, & DeShon, 2003). Moreover, research on both service climate and safety climate has been systematic, developing a set of findings that convincingly demonstrate significant effects on customer perceptions of service quality (Schneider & Bowen, 1985; Schneider, White, & Paul, 1998) and increases in safety-related behaviors with corresponding reductions in objective accident indicators, respectively (Hofmann & Stetzner, 1996; Zohar, 2000, 2002; Zohar & Luria, 2004). Schneider et al. (1998), for example, showed that organizational policies and practices relating to a strategic imperative for service influenced shared employee perceptions of the service climate, which in turn influenced customer perceptions of service quality. As research supporting the utility of collective climates for predicting meaningful organizational, unit, and individual outcomes began to accrue, researchers also began to explore factors that influenced perceptual consensus and the emergence of unit climate. Harkening back to the roots of climate theory (Lewin et al., 1939), Kozlowski and Doherty (1989) proposed that leaders shape the interpretation of climate for those team members with whom they have a good leader-member exchange (LMX; Graen, Orris, & Johnson, 1973) relationship. Their results showed that team members with good LMX relations had climate perceptions that were both similar to their leader and consensual with each other, relative to those with poor LMX relations, whose perceptions were discordant with the leader and each other. More recent research has provided results consistent with this, showing that team leaders and the quality of their LMX relationships they enact with team members play a key role in shaping the nature and strength of climate perceptions (Hofmann & Morgeson, 1999; Hofmann, Morgeson, & Gerras, 2003; Zohar, 2002; Zohar & Luria, 2004). Rentsch (1990), building on Schneider and Reichers' (1983) suggestion that social interaction shapes consensual climate, showed that networks of individuals with frequent informal social interactions showed greater consensus on climate relative to consensus within those individuals' formal organizational units. This research demonstrates nicely the theoretical assertion that social interaction contributes to shared climates. In an effort to better capture climate emergence, Brown, Kozlowski, and Hattrup (1996) suggested that sharing, consensus, or agreement on climate perceptions ought to be a substantive phenomenon of interest rather than a mere statistical justification for aggregating individual climate perceptions to represent unit climates. Rather than treating the emergence of climate as all or none (i.e., if agreement is above threshold, climate perceptions are aggregated; if it is below threshold, there is no support for aggregation), they argued that degree of perceptual agreement could be conceptualized as measuring the extent to which climate perceptions were dispersed, had started to converge, or were highly consensual within the unit and therefore the extent to which climate had emerged. On this basis, Chan (1998) and Brown and Kozlowski (1999) proposed dispersion models in which perceptual agreement indexed the strength or degree to which a perceptual construct, in this case climate, emerged at a higher level of analysis (Kozlowski, 1999). Subsequent work has generated support for this dispersion-theory conceptualization. For example, Schneider, Salvaggio, and Subirats (2002) reported data showing that aggregated (unit means) individual-level employee perceptions of unit (bank branches, N = 118) service climate significantly interacted with unit climate strength (unit standard deviation indexing the degree of within-unit consensus), thereby demonstrating that greater within-unit consensus or strength yielded stronger relationships between service climate and aggregated customer perceptions of service quality concurrently and over a 3-year time interval. Examining data from 197 work units, Gonzlez-Rom et al. (2002) showed that social interaction and leader-informing behavior was positively related to climate strength within units. Moreover, climate strength interacted with aggregated innovation climate to influence average unit satisfaction and commitment, and climate strength interacted with aggregated goal orientation to influence average unit commitment. Based on this systematic body of theory development and empirical support, we conclude that a collective climate that captures the strategic imperatives reflective of the core mission and objectives of an organization, unit, or team is a key emergent cognitive structure that shapes processes relevant to goals and their accomplishment. We further conclude that factors influencing climate consensus or strengthsuch as strategic imperatives, leadership, and social interactionrepresent leverage points for shaping collective climates to influence team effectiveness. Team Mental Models and Transactive Memory Team mental models and transactive memory both refer to cognitive structures or knowledge representations that enable team members to organize and acquire information necessary to anticipate and execute actions. As we detail below, team mental models refer to knowledge structures or information held in common, whereas transactive memory refers to knowledge of information distribution within a team (i.e., knowledge of who knows what). Although these concepts may seem to bear some superficial resemblance to team climate, they are quite distinctive. Where climate tends to be more general in nature, team mental models and transactive memory are more specific to the team task and work system. Where climate is about what should be aimed for and, perhaps, why, team mental models and transactive memory are about how the knowledge to do it is organized, represented, and distributed. Team mental models capture the shared, organized understanding and mental representation of knowledge or beliefs relevant to key elements of the team's task environment (Klimoski & Mohammed, 1994). The concept of a mental model developed in the human-factors literature as an expert's cognitive representation of a system that could be used for predicting system states and for generating inferences about system behavior (Rouse & Morris, 1986). Cannon-Bowers, Salas, and Converse (1993) posited that such cognitive representations, held in common by team members, might help enable team members to anticipate needs and actions and thereby "implicitly" coordinate their behavior and improve team effectiveness. We should highlight that the enhancement of coordination has been posed in the context of action teams that perform critical tasks under dynamic uncertainty (e.g., cockpit crews, surgical teams, command and control, combat teams), where coordination is a critical team process. This speculation energized considerable interest and activity in the team-research community. Four primary content domains of team mental models were originally proposed (Cannon-Bowers et al., 1993). They include (a) knowledge about the equipment and tools used by the team (equipment model); (b) understanding of the team task, including its goals, performance requirements, and problems (task mental model); (c) awareness of team-member composition and resources, including representations of what individual members know and believe and their skills, preferences, and habits (team-member model); and (d) what team members know or believe about appropriate or effective processes (team-interaction model or teamwork schema). One of the biggest challenges in research and application centers on how to measure and represent team mental models as group-level cognitive structures (Mohammed, Klimoski, & Rentsch, 2000). Mohammed et al. (2000) identified four techniques, including pathfinder networks (PF; which generates a node and link structural representation based on ratings of the psychological proximity among concepts), multidimensional scaling (MDS; which generates a representation in geometric space based on ratings of the psychological proximity of concepts), interactively elicited cause mapping (IECM; which characterizes the causal linkage among concepts based on observation, interviews, and questionnaire data), and text-based cause mapping (TBCM; which characterizes the causal linkage among concepts based on text-based input), that can be useful for measuring team-level cognitive structure. The evaluation of the strengths and weaknesses of these different techniques by Mohammed et al. (2000) was based on seven criteria including the treatment of content, evaluation of structure, a clear standard, reliability evidence, effectiveness, utility for team analyses, and other considerations. They concluded that the decision of which technique to use depends on the research question under examination and on the team context. For example, if researchers know the key concepts of a domain and want to capture the structural relations among concepts (e.g., to compare expert and novice mental model structures), then PF and MDS are appropriate because they provide tools to represent, describe, and compare structures. On the other hand, if researchers are interested in measuring team members' belief relations, then IECM and TBCM may be more appropriate because they necessitate the direct elicitation from participants of concepts and relational information that capture the richness of cognitive content. Researchers also have to be mindful of levels-of-analysis concerns when data are gathered from individuals but when the goal is to represent the team level (Kozlowski & Klein, 2000). The presumption of the shared-mental-model literature is that team effectiveness will improve if team members have a shared understanding of the task, team, equipment, and situation (Cannon-Bowers et al., 1993). Although the notion of a common or shared team mental model has tended to dominate the research, there is recognition that team mental models may be more complex, in the sense that members do not necessarily have isomorphic (i.e., identical) knowledge structures but possess some sharing and also some unique structural information (e.g., based on role distinctions) that is compatible with that of other member roles (e.g., Banks & Millward, 2000; Kozlowski, Gully, Salas, & Cannon-Bowers, 1996). This conceptualization of networked knowledge incorporates aspects of a distributional model somewhat similar to the notion of transactive memory. However, most empirical work has focused on shared knowledge organization. Although there have been some concerns regarding the adequacy of empirical work relative to conceptual development (Mohammed & Dumville, 2001), research has accumulated to provide solid support for the general presumption that shared mental models are associated with team effectiveness. Minionis, Zaccaro, and Perez (1995), for instance, used concept maps to examine shared mental models among team members in a computer-based, low-fidelity tank simulation. Their results indicated that shared mental models enhanced performance on collective tasks requiring interdependence among team members but did not impact those tasks that could be completed without coordinated actionswhich is consistent with the general thesis. Mathieu, Heffner, Goodwin, Salas, and Cannon-Bowers (2000) examined the effect of shared mental models on team processes and performance using two-person teams performing a PC-based combat flight simulation. Their results indicated that both teamwork and taskwork mental models related positively to team process and performance and that team processes fully mediated the relationship between shared mental models and performance. There has also been theory and research to identify interventions that enhance the development of team mental models. Focusing on leadership, Kozlowski and colleagues (Kozlowski, Gully, McHugh et al., 1996; Kozlowski, Gully, Salas et al., 1996) posited that team leaders could play a key role in shaping team mental models by linking task cycles or episodes to a regulated learning process. Prior to action, the leader helps set team learning goals commensurate with current team capabilities; during action, the leader monitors team performance (and intervenes as necessary); and as the team disengages from action, the leader diagnoses performance deficiencies and guides process feedback. This cycle iterates and the leader increments the complexity of learning goals as team skills develop and compile. Marks, Zaccaro, and Mathieu (2000) showed that leader pre-briefs that focused on appropriate team task strategies yielded better team mental models, processes, and performance. Other research focusing on the effects of leader pre-briefs (planning, strategies) and debriefs (feedback) has also shown positive effects on team mental models and team performance (Smith-Jentsch, Zeisig, Acton, & McPherson, 1998; Stout, Cannon-Bowers, Salas, & Milanovich, 1999). Planning, strategies, and contingencies enable teams to overcome obstacles to goal accomplishment (Tesluk & Mathieu, 1999), and training (described below) that focuses on skills relevant to diagnosis, feedback, and planning is also important (Blickensderfer, Cannon-Bowers, & Salas, 1997). As a set, this research suggests that team leaders can play a central role in shaping the formation of shared mental models (as well as in team learning; Edmondson, 1999). In addition, there are several training techniques that are effective for shaping the development of team mental models and enhancing team performance. For example, using a low-fidelity tank simulation, Marks et al. (2000) employed a concept-mapping technique to assess team mental models and showed that mental-model formation was influenced by leader pre-briefs and team-interaction training. Mental models, in turn, influenced team-communication processes and performance. Moreover, the effects of team mental models and communication processes on team performance were stronger when the task conditions were novel, suggesting that team mental models may help team members adapt to the unexpected. Marks, Sabella, Burke, and Zaccaro (2002) examined the effects of cross-training (i.e., training on other team-member responsibilities in an interdependent task) on team-interaction mental models and their subsequent effects on coordination, back-up behavior (i.e., helping a teammate), and performance. Their results indicated that cross-training improved team-interaction mental models, assessed via PF, and that the effects of team mental models on team performance were mediated by improved coordination and back-up behavior. Blickensderfer et al. (1997) developed the technique of team self-correction training as a means to enhance the natural mechanism by which team members correct their team attitudes, behaviors, and cognitions. Team self-correction training focuses on skills relevant to (a) event review (following a task episode), (b) error identification, (c) feedback exchange, and (d) planning for subsequent task episodes. Much of this growing empirical support for the efficacy of shared team mental models for fostering team processes and performance has been conducted in the context of action or command-and-control teamsreal or simulatedbut there is evidence in more recent work that it extends to other types of teams, such as TMTs (e.g., Ensley & Pearce, 2001; Knight et al., 1999). On the basis of this systematic body of theory development and growing empirical support, we conclude that a shared team mental model that captures the structure of relations among key aspects of the team, its task and role system, and its environment is a key emergent cognitive structure that shapes coordination processes relevant to team goals and their accomplishment. We conclude that factors influencing the development of shared team mental modelssuch as leadership, training, and common experiencerepresent leverage points for shaping the formation of team mental models that influence team effectiveness. The concept of transactive memory, as a memory system distributed across group members, was first proposed by Wegner to explain why close personal relationships often foster the development of common memory (Wegner, Giuliano, & Hertel, 1985), and much of the formative work on transactive memory has studied it in the context of intimate relationships (e.g., dyads in dating relationships). From a team perspective, transactive memory is a group-level collective system for encoding, storing, and retrieving information that is distributed across group members (Wegner, 1986, 1995; Wegner et al., 1985). In contrast to the more uniform knowledge sharing characteristic of shared mental models, transactive memory is conceptualized as a set of distributed, individual memory systems that combines the knowledge possessed by particular members with shared awareness of who knows what (Wegner, 1995). In that sense, one can conceptualize transactive memory as a network, with team members and their unique knowledge representing nodes with links representing other members' awareness of that unique knowledge. When each team member learns in a general sense what other team members know in detail, the team can draw on the detailed knowledge distributed across members of the collective. The development of transactive memory involves the communication and updating of information members have about the areas of the other members' unique knowledge. Each member keeps track of other members' expertise, directs new information to the matching member, and uses that tracking to access needed information (Mohammed & Dumville, 2001; Wegner 1986, 1995). In this way, team members use each other as external memory aids, thereby creating a compatible and distributed memory system. Given the presumed distribution of specialized memories across team members, transactive memory systems should be more cognitively efficient, allowing high specialization and greater capacity. Conceptually, such systems should reduce the cognitive load on individuals, enlarge the collective pool of expertise, and minimize redundancy (Hollingshead, 1998b). On the other hand, there are likely to be limits to the size of such a distributed memory system, at which point tracking costs may outweigh memory gains. There are important concerns that relate to conflicts of expertise, failure to capture important information, and diffusion of responsibility (Wegner, 1986). And, there are time lags and efficiency costs for accessing a distributed memory that may be detrimental to team effectiveness in time-critical situations (Kozlowski & Bell, 2003). Transactive memory is an intriguing concept that augments the focus on shared knowledge that dominates the team-mental-model literature. Unfortunately, however, empirical research on transactive memory is not commensurate with its theoretical development. Moreover, compared to the research base on team mental models, research on team transactive memory is still in its infancy. In an early application by Liang, Moreland, and Argote (1995), undergraduates were trained to assemble a radio either individually or in groups that were later tested with their original group or in a newly formed group that was composed by mixing members from different groups. Members of groups that had trained together evidenced stronger transactive-memory systems specialized for remembering different aspects of the task, they coordinated more effectively, and they showed greater trust in other members' expertise. In addition, transactive memory mediated the effects of group training on task performance. A follow-up study conducted by Moreland (1999) used a similar task and design. He used a more direct measure of transactive memory by assessing the complexity of group members' beliefs about other members' expertise, belief accuracy, and agreement about the distribution of expertise within the group. Rulke & Rau (2000) used qualitative methods to examine the development of transactive memory, examining the sequence by which encoding unfolds over time. More recent research has begun to address some of the conceptual and measurement problems identified by Kozlowski and Bell (2003) that have plagued this literature. Austin (2003), for example, integrated prior research to develop assessments of four dimensions of transactive memory, including (a) knowledge stock as the pool or combination of individual knowledge, (b) consensus about knowledge sources, (c) specialization of knowledge, and (d) the accuracy of knowledge identification. Examining 27 product teams (N = 263) in a sporting-goods company, results indicated that both specialization and accuracy were substantially related to external and internal evaluations of effectiveness and that accuracy was also related to external ratings of goal accomplishment. Lewis (2003), drawing on Moreland (1999), developed a three-dimensional perceptual measure of transactive memory. The dimensions included specialization, credibility, and coordinationwhich, although labeled differently, are consistent with Moreland (1999) and Austin (2003). Notably, within-group agreement on the transactive memory perceptions was used as a justification for aggregation of the perceptions to the team level. The study was rigorous in its development and validation of the assessment method. A related paper that appears to be based on a portion of the same data (Lewis, 2004) concluded that frequent face-to-face communication (but not other forms of communication) facilitated the formation of transactive-memory systems and that transactive memory was related to team performance and viability. This recent research is promising, although there is also a need for clear conceptual and empirical demarcations across the team cognitive constructs of climate, mental models, and transactive memory. For example, the assessment approach developed by Lewis (2003), which necessitates within-group perceptual consensus, blurs the distinction between climate, team mental models, and transactive-memory systems. The content of its perceptual dimensions is based on transactive-memory conceptions, but because it incorporates the assumption of shared within-team perceptions to support data aggregation it mirrors assessments of shared team climate and some assessments of shared team mental models. It is the distribution of unique information that makes transactive memory distinctive. This observation is not meant to suggest that the technique is flawed. Rather, we simply note the need for clear conceptual and empirical demarcations across these team cognitive constructs. There is merit to both the team-mental-models and transactive-memory approaches, but there are also important conceptual distinctions that may be differentially important for teams under different contingencies. In particular, it is likely that the difference between task interdependence, which necessitates differentiated role specialization (i.e., distributed expertise), and pooled tasks, which necessitate more common knowledge, will bear on the importance of knowledge distribution that would make transactive memory more important relative to knowledge sharing that would make shared mental models more relevant (Hollingshead, 2001). As research on transactive memory goes forward, there is a need to expand on the formative work (e.g., Austin, 2003; Lewis, 2003, 2004) that examined its formation, emergence, and effects in work teams (Mohammed & Dumville, 2001). With respect to shaping the formation of transactive memory, some research suggests that face-to-face interaction is important for its emergence and that computer-mediated communication presents barriers (Hollingshead, 1998a; Lewis, 2003). Given the rise of virtual teams, this finding should prompt concerted efforts to understand the reasons for such barriers and to try to develop augmentations for computer-based and other forms of remote communication. Research also shows that common experience or training (i.e., learning the task together as an intact team) may be useful for developing transactive-memory systems (Liang et al., 1995; Moreland, 1999). Thus, this may be an area in which the use of interpositional cross-training, which has proven useful in the development of shared mental models, may also help to foster the development of better transactive-memory systems. In summary, we believe that the concept of transactive memory shows considerable promise for explaining distributed and compatible memory and knowledge systems in teams, making it a unique and useful supplement to the concept of team mental models. On that basis, we conclude that teams with better transactive-memory systems (i.e., knowledge of member specialization and strategies to access the knowledge) will be more effective. However, beyond familiarity, shared experience, and face-to-face interaction, the research base to help identify techniques for enhancing transactive memory is as yet not sufficiently developed to warrant specific recommendations for how to enhance it in teams. This is an obvious target for vigorous and rigorous research. Team Learning The concept of group or team learning refers to the acquisition of knowledge, skills, and performance capabilities of an interdependent set of individuals through interaction and experience. Team learning is fundamentally based on individual learning, but when viewed as more than a mere pooling of individual knowledge it can be distinguished as a team-level property that captures the collective knowledge pool, potential synergies among team members, and unique individual contributions. Certainly, the concept of learning is broad enough to be applied to levels of social organization beyond the individuali.e., to teams, organizations, and collections of organizationsand it has been so applied. Indeed, the literature on organizational learning has developed over the last four decades into a rich, multifaceted, and multidisciplinary area of inquiry focused on creating, retaining, and transferring knowledge among higher-level entities (Argote, McEvily, & Reagans, 2003). One of the key challenges is to distinguish team learning from related concepts such as team mental models and transactive memory, which also develop through collective interaction and common experience. One way to make such a distinction is to regard team climate, mental models, and transactive memory as emergent states that develop from learning as a dynamic behavioral process of interaction and exchange among team members (Kozlowski & Bell, in press). When viewed as a process, it becomes more apparent that learning is contextually based and socially bound. For example, the concept of vicarious learning represents a potent form of knowledge transfer within groups, and it is certainly an important means by which knowledge is transmitted and acquired in group contexts (Argote et al., 2003). One emerging perspective on collective learning that is consistent with this process view is represented by Edmondson's (1999) model of team learning. She noted that much of the relatively limited literature on team learning has been conducted in the laboratory, which limits the nature of the phenomena that can be observed. Edmondson provided a rigorous evaluation of her model in 51 work teams, using a combination of qualitative and quantitative techniques. The key construct in this model is the team perception of psychological safety, a climate-like shared perception that the team is a safe context for interpersonal risk taking. Team psychological safety, in turn, influenced team learning behaviors indicative of a team learning processsuch as seeking feedback, sharing information, experimenting, asking for help, and discussing errorswhich then influenced team performance. Finally, a supportive organizational context and effective coaching by the team leader contributed to the development of perceptions of team psychological safety (see also Edmondson, Bohmer, & Pisano, 2001). Although Edmondson's work focuses on the learning aspects of these team behaviors, in a subsequent section we will address similar team behaviors as underpinnings for team performance as a dynamic process. Because learning, motivation, and performance are entwined, Edmondson's approach to team learning provides a useful linkage with this other emerging work, suggesting the potential for an effective integration in a broader model that will become apparent in the concluding section of this monograph. Apart from Edmondson's research, there has been relatively little research on team learning outside of the laboratory. Here we provide a couple of illustrations of lab research on team learning. For example, Argote, Insko, Yovetich, and Romero (1995) studied the impact of individual turnover and task complexity on learning in a laboratory group task. Group performancemaking origami birdsexhibited a performance learning curve, with output increasing significantly at a decreasing rate over six trials. High turnover and high complexity of the task were detrimental to group performance, with the detrimental effect increasing as groups gained experience, suggesting in the case of turnover that aspects of group knowledge were being lost. Using a complex command-and-control radar simulation and four-person teams, Ellis et al. (2003) found that team composition and structure influenced learning. Team learning was assessed as an aggregate of the effectiveness and efficiency with which individual team members took action against unknown radar contacts, where effectiveness and efficiency represented the degree of match between resources expended to take action and the capabilities possessed by a contact (e.g., if a contact was moderately powerful and the action taken used excessive power, the action would be considered inefficient). Because each member had some unique information to help identify the type of contact, teams that were more effective and efficient on average could be inferred to have shared information or learned from one another. Teams composed of members with higher average cognitive ability and teams whose workloads were more evenly distributed among members learned more. With respect to personality composition, those teams whose members were higher on agreeableness did worse. Finally, teams configured as two subgroups learned more than did those without subgroups. Although there is interesting work being conducted on collective knowledge management, it is largely at levels beyond teams and focuses on processes that are more macro in origin and effect (e.g., Argote et al., 2003; Epple, Argote, & Devadas, 1991). At best, the research on team learning is still in the formative stage and needs improved focus and rigor. First, team learning as an outcome is typically inferred from changes in team performance. It is rarely assessed directly as a construct in its own right. Relatedly, many of the factors having an impact on team learning (e.g., turnover) are also likely to affect team performance via pathways other than learning. That is, while the removal or replacement of team members can influence collective knowledge, changing team members will also disrupt communication, coordination, and other emergent states that directly affect team performance. It is impossible to disentangle the process pathways and to map the pattern of effects as long as researchers continue to rely on the assumption that teams have learned rather than on direct assessments of team learning. Thus, there is a need for research in this area to directly measure changes in individual and team knowledge or some other direct evidence that learning has occurreddistinct from behaviors that directly contribute to performance. In a related vein, there is some question as to whether team learning as a knowledge-based outcome can be meaningfully distinguished from team mental models and transactive memory. In essence, one could consider team mental models and transactive memory as different manifestations of team learning, but then team learning would have no unique standing as a construct. It would merely be a more general conceptual handle for referencing more specific constructs. Second, as we suggested earlier, it may be more valuable to conceptualize team learning as a process that yields shared knowledge, team mental models, and transactive-memory structures as emergent states (Kozlowski & Bell, in press). However, with the exception of the work conducted by Edmondson (1999), there has been relatively little research to specify the process by which team learning occurs. Recent theoretical work by Ellis and Bell (2005) that conceptualizes team learning as a form of information processing (Hinsz et al., 1997) involving key conditions of capacity, collaboration, and commonality may also be useful in elaborating a process view. A process perspective necessitates simultaneous attention to both the individual and team levels of analysis. Researchers would not be able to simply combine disparate individual actions into an aggregate. Rather, they would have to more carefully delineate how disparate individual learning and action emerge as a collective phenomenon. We think that such an effort has long-term potential to illuminate the process of collective learning. In summary, we see team learning as another team-level cognitive resource that has promise to help in understanding how team members are able to combine their knowledge to improve team effectiveness. On that basis, we conclude that teams that learn more collectively will demonstrate enhanced effectiveness. However, we also conclude that the research base to specify the meaning of team learning as a construct distinct from other team cognitive constructs and emergent states, and the research base to identify antecedents that enhance team learning, are not yet sufficiently developed to warrant specific recommendations for how to enhance team learning beyond individual knowledge acquisition. This is an obvious target for vigorous and rigorous theory development and research. Team Interpersonal, Motivational, and Affective Processes and Emergent States Research on processes has a long history in small-group research, with much of that effort traditionally centered on processes that capture motivational tendencies, relations among team members, and affective reactions. These processes imply a dynamic unfolding, but measures of team processes are most often static assessments that fail to capture temporal dynamics. That is why Marks et al. (2001) suggest that it is more accurate to describe the construct measures of team processes as emergent states. We will use both terms interchangeably and explicitly acknowledge that these constructs emerge from dynamic interactions among team members that tend to stabilize over time. We focus on a set of constructsteam cohesion; team efficacy and group potency; affect, mood, and emotion; and team conflictthat capture bonding to the team and its task; confidence in members' task competencies; affective processes and reactions; and their fractures, frictions, and disagreements. Team Cohesion Group cohesion is one of the earliest and most widely studied team-process characteristics. Team researchers have offered multiple definitions of cohesion. Festinger (1950) defined cohesiveness as "the resultant of all the forces acting on the members to remain in the group" (p. 274). Festinger is also responsible for suggesting the three facets that comprise cohesionmember attraction, group activities (i.e., task commitment), and prestige or group pridealthough early researchers tended to view cohesion as a unitary construct. Nonetheless, other researchers have often emphasized only one of the facets. For example, Evans and Jarvis (1980) identified mutual attraction of members to the collective as the most common definition of cohesiveness. Carron (1982) defined cohesiveness as a process that reflects a group's tendency to stick together and remain united to reach a common goal. Goodman, Ravlin, and Schminke (1987) defined cohesiveness as the commitment of members to the group's task. These somewhat mixed definitions or emphases are reflected in different conc.

Summarize this article