Conflict in Diverse Teams: The Problem of Representational Gaps And the Solution of Cognitive and Affective Integration

Diversity in teams, while potentially beneficial, increases the likelihood that individual team members will define the team’s task differently, leading to gaps in their perceptions of what is important for team task execution. These representational gaps are likely to create conflict as teammates try to solve what are essentially incompatible problems. We articulate how these representational gaps affect team functioning and how cognitive and affective integration can mitigate the potentially negative effect of representational gaps. REPRESENTATIONAL GAPS AND CONFLICT 3 Organizations often rely on functionally diverse teams to work on complex problems: top management teams often include representatives from finance, marketing, and operations; product development teams include people from product design, engineering, marketing, and manufacturing. Members of functionally diverse teams bring unique perspectives to the task at hand which can lead to creative or innovative thinking (Fiol, 1995; Galbraith, 1977; Milliken & Martins, 1996; Williams & O'Reilly, 1998). However, different perspectives can lead team members to have different views of the team’s ultimate task (Dearborn & Simon, 1958; Dougherty, 1992). When these views are inconsistent we say there are “representational gaps” between individual team member’s cognitive problem representations. When there are representational gaps, we argue that conflict over what to do to solve the problem (i.e. task conflict) is likely, and will also be difficult to manage. Task conflict is more likely to interfere with than facilitate performance (De Dreu & Weingart, 2003b), and so we argue that it is important to mitigate the representational gaps. To this end we also discuss two means for managing representational gap conflict: cognitive integration and affective integration. Cognitive integration is defined as the degree to which people understand each others’ perspective (e.g., the assumptions they make, the issues they are concerned with) and should reduce conflict through reducing the effect of diversity on representational gaps. Affective integration is defined as the degree of trust, respect, and liking teammates have for one another, and should reduce conflict by moderating the effect of representational gaps on conflict. Our model is presented in Figure 1. ----------------------------------------------Insert Figure 1 about here ----------------------------------------------REPRESENTATIONAL GAPS AND CONFLICT 4 A representational gap is defined as an incompatibility between team members’ cognitive representations of the problem. Because a representational gap occurs in the conceptualization of the problem, it represents more than alternate views about what to do, it represents alternate views about what the problem is. For example, at a particular auto manufacturer there is a strong focus on quality in their product development teams; it is their top priority. Yet if you ask their engineers what quality implies, they will say that a quality product works the way it is supposed to, is reliable and durable. Designers, on the other hand, will say that a quality product looks finely crafted, is made of good material, and has pleasing contours. These different interpretations of “quality” demonstrate a gap between designers’ and engineers’ definitions of the same task – to make a quality vehicle. Differences among team members’ views of the same task have been addressed in the research on shared mental models (Cannon-Bowers, Salas, & Converse, 1993; Klimoski & Mohammed, 1994; Rouse & Morris, 1986). However, in contrast to the shared mental model perspective, we argue that it is the compatibility between and not the similarity of team members’ views of the problem that matters for team functioning, hence our focus on gaps and not overlap. A focus on compatibility allows us to recognize that the differences between mental models are informative, not just the similarities. For example, that engineers in a multi functional team see their problem in terms of weight, cost, and functionality while designers see their problem in terms of vehicle theme, brand identity, and innovative styling is not a cause for concern because their perspectives are (potentially) compatible – these interests can be met simultaneously. However, these same team members do not hold a shared mental model of vehicle value and quality. Considering compatibility of problem conceptualizations more accurately reflects our experience with multi-functional team success factors. These teams are REPRESENTATIONAL GAPS AND CONFLICT 5 purposely specialized and coordinated, having little overlap in how the functional perspectives view the team’s task. Specialization is adaptive in part because a problem is too complex for one person to understand in its entirety, and in part because the interaction between different perspectives occasions creativity (Amabile, 1988; Milliken et al., 1996; Perry-Smith & Shalley, 2003; Williams & Yang, 1999). Focusing on compatible knowledge rather than common knowledge provides a theory that aligns with a set of findings that defy the shared mental model claim that performance improves as the team members’ mental models increase in similarity (e.g., Cannon-Bowers et al., 1993; Espinosa & Carley, 2001). One such example is that many well functioning teams have highly differentiated roles and minimal overlap within the group (see Hutchins, 1990). In addition, transactive memory research (Argote, Gruenfeld, & Naquin, 2001; Liang, Moreland, & Argote, 1995; Moreland, Argote, & Krishnan, 1998; Wegner, 1987) has shown that teams with specialization of knowledge, where there is less common knowledge, were superior to those without such differentiated knowledge. Finally, Levesque, Wilson & Wholey (2001) found that successful software development teams’ mental models diverged over time. All of these findings run counter to the simple notion that the more shared knowledge among group members, the better. To be fair, some have suggested that there is likely to be an optimal degree of sharing that falls short of complete overlap (Cannon-Bowers et al., 1993; Mohammed, Klimoski, & Rentsch, 2000). However, the shared mental model literature has yet to say either what this is or how it should be determined. By focusing on representational gaps, we believe we can get closer to answering this question (which we do later in the paper). At one level, the purpose of this paper is to show the value of refocusing attention from what is shared between teammate’s views of the problem to the gaps between those views. REPRESENTATIONAL GAPS AND CONFLICT 6 However, the primary aim of the paper is to understand how diverse teams can be negatively influenced by the existence of representational gaps, and what should be done in order to ameliorate these problems and thereby maximize the potential of these teams. The ability to bridge the gaps requires an understanding of the nature of representational gaps and the conflicts that they create. By understanding the mechanics of representational gap conflict we seek to contribute to theories of team diversity by explicating the cognitive and affective processes through which diversity can help (or hinder) performance and creativity. In addition we seek to deepen our understanding of conflict by explicating a cause at the root of many conflicts in multi functional teams. In this paper, we begin by specifying what representational gaps are, and why team diversity makes them more likely. We then examine the consequences of representational gaps on conflict and by extension team functioning. We introduce cognitive and affective integration as ways to deal with representational gaps in order to reduce conflict. We conclude by examining the implications for diverse teams, and then consider other situations to which the concepts generalize. Our theory is grounded in observation of multi functional teams in organizations. Throughout the paper we will draw on examples from one firm in particular, GiPSE, a large vehicle design and manufacturing company, who used multi functional teams to increase creativity and efficiency, but experienced many of the problems we will detail. COGNITION IN GROUP PROBLEM SOLVING In order to understand how representational gaps lead to conflict, we must understand what a representational gap is in the context of problem-solving by interdependent actors. We develop this theory of joint problem solving by considering the relevant individual level REPRESENTATIONAL GAPS AND CONFLICT 7 mechanisms and then theorizing how they link together when more than one person attempts to solve a problem. Problem solving is the process of applying knowledge to the current situation in order to achieve desired objectives (Hayes & Simon, 1974; Newell & Simon, 1972). Generating relevant knowledge requires a cognitive model of the problem to process new information and guide retrieval of existing knowledge; this is called the problem representation (Chi, Feltovich, & Glaser, 1981; Hayes et al., 1974; Newell et al., 1972; Simon, 1979). The representation is “a cognitive structure that corresponds to a given problem... constructed by the solver on the basis of domain-related knowledge and its organization” (Chi et al., 1981, p. 131). The representation is comprised of the goal hierarchy, assumptions, elements, and operators (GAEO) (Hayes & Simon, 1974; Newell & Simon, 1972); these provide the means for solving the problem. The goal hierarchy specifies the precedence and relative importance of the objectives that need to be met in order to solve the problem. For example, the goal hierarchy for building an automobile could include two goals: affordability and innovativeness in styling, and affordability may be more important than styling by a wide margin. Goals activate the associated knowledge in memory and direct people’s attention to similarly relevant knowledge in the environment (Anderson, 1993). The assumptions are the “givens” of the problem. They can take the form of postulates (e.g. people act in their own best interest) or restrictions (e.g. the project must be completed in 10 days). The assumptions determine what actions/ outcomes people believe to be possible as well as desirable. In essence, assumptions bound the region where people will search for knowledge and limit what people will try to change. The elements are the components of the problem that are changeable. In a physical problem it is easy to think of elements as the “moving parts”; however, one could consider abstract constructs as elements as well (e.g., vehicle theme, REPRESENTATIONAL GAPS AND CONFLICT 8 profit margin). A person applies operators to the elements in order to reconfigure the elements. As one applies operators to the elements (i.e., makes a move), the problem state should look more and more like the desired end (the goal state). It is critical to recognize that the representation directs problem solving by guiding what is believed to be possible and desirable. The elements direct attention to what can be altered, and the operators underlie how these things can be altered, thereby generating the candidate moves at any given point. The assumptions check the candidate moves to make sure they have not violated any rules. The goal hierarchy provides the basis for evaluating moves against each other, as a move is preferable in direct relationship to how much it satisfies (or brings one closer to satisfying) the goals. It is also critical to realize that representations are simplifications of the task environment (Newell & Simon, 1972). Representations economize on attention; they direct people to a subset of the details in a complex task environment. Representations are also subjective in that different people can construct different cognitive models of the same problem. Joint-representations. Often, however, individuals work in conjunction with others on a shared task. In this situation, an individual will need to consider their own role within the context of the group and thus will create a joint-representation. A joint-representation coordinates team action by integrating one’s own GAEO with the relevant GAEO of others, thus, it positions one’s own problem in the “bigger picture” of the group’s problem. At some level, a joint-representation is analogous to a shared mental model (CannonBowers et al., 1993; Klimoski et al., 1994; Rouse et al., 1986). Both address the way an individual cognitively represents a team’s task. As a construct, we use “joint-representation” rather than “shared mental model”. Joint-representation as a construct does not have the same problems of definitional ambiguity as does “shared mental model”. Ten years ago, Klimoski and REPRESENTATIONAL GAPS AND CONFLICT 9 Mohammed (1994) entreated researchers to be more rigorous in their conceptualization in order to work toward consensus on a generally accepted definition of shared mental model. Unfortunately, the definitions seem to have diverged and not converged (Mohammed et al., 2000). In contrast, joint-representation is a small and logical expansion of the original concept of individual problem representation, a concept with a long consistent history in information processing theory (Newell, 1990; Newell et al., 1972). In addition to problems with consensus on the overall construct, there is little agreement on either the definitions of the sub-dimensions of shared mental model (e.g., team mental model, task mental model, tool mental model), or how many different subtypes of mental models there are (Mohammed et al., 2000, mention there are at least four). It is unclear how separable these mental model subtypes are (does team mental model include team member abilities?) either theoretically or in people’s minds. In contrast, joint representation is decomposable into the GAEO, well defined, mutually exclusive and exhaustive components (Hayes et al., 1974). Perhaps the best reason to use joint representation rather than shared mental model is that the joint representation concept can provide direction on the question posed but not answered in much of the shared mental model literature (e.g., Cannon-Bowers et al., 1993; Mohammed et al., 2000): What needs to be shared among team members to have a well functioning team? One can make specific predictions based on the team’s function using the GAEO. Groups that require their members to be interchangeable, for example, infantrymen in a particular unit of the military, would require identical joint representations where everyone shares the same GAEO (Figure 2). In these teams, any individual may be called upon to do any of the jobs of the other individuals, and because the groups must be able to adapt quickly to changes in the environment, they must all have similar skills and interpretive frameworks. Groups with a high degree of REPRESENTATIONAL GAPS AND CONFLICT 10 specialization and coordination, such as multi-functional teams, will have more modular joint representations. In a modular joint representation (Figure 3), only the goal hierarchy is completely shared, as all must be working toward the same end. The assumptions are shared between only those whom they may restrict (e.g., I only have to worry about your assumptions about weight when it affects my choice of material). In these teams, because the task is complicated but fairly static, it makes more sense to have experts in each of the relevant areas working in tandem on the overall task. Full exploration of the relationship between task parameters and required overlap in GAEO is beyond the scope of this paper, suffice to say that joint representation, unlike shared mental model, provides well defined concepts that should make it easier to diagnose how much shared knowledge is sufficient. ----------------------------------------------Insert Figures 2and 3 about here ----------------------------------------------The joint-representation coordinates member actions by positioning an individual’s GAEO in relation to those of the other team members. Positioning the individual’s goals within the group’s goal hierarchy should provide the individual with an understanding of how important his or her individual goals are in relation to the other’s individual goals (e.g., is it more important to meet my deadline or help you meet yours). It should also position one’s individual goals in relation to the overall group goals (e.g. how much of a limited budget to give to product design when product appearance is not the group’s most important goal). A shared goal hierarchy is critical for coordinating the use of shared resources, as well for directing how the individual devotes resources to his or her task(s). Another way the joint-representation coordinates action is through shared assumptions. The assumptions of one teammate should not counter the REPRESENTATIONAL GAPS AND CONFLICT 11 assumptions of another, as this would lead teammates to different conclusions about the plausibility of a move. In terms of elements and operators, a joint-representation serves to both expand and restrict the particular moves an individual makes. A joint representation that includes others’ operators can expand the possible moves that person can make. For example, when collaborating with a statistician, a psychologist may learn of many new analysis techniques the statistician can perform (i.e., statistician’s operators) on new data types (i.e., elements). This may alter the data collection techniques employed in executing the experiment (psychologist’s moves). A joint representation should also serve to restrict how operators are used in relation to others assumptions. This is important for move selection because parties must not execute operators that change the problem into an illegal (i.e., unacceptable) state for another party (e.g. engineering should not order a material that exceeds finance’s budget restriction on that part). REPRESENTATIONAL GAPS AND THOUGHT WORLD DIVERSITY A representational gap is an incompatibility between team members’ jointrepresentations. Specifically, a representational gap manifests as a contradiction (e.g., one person thinks the group’s product should value design above cost, the other believes the reverse) or omission (e.g., one person assumes that the deadlines are fixed, a second does not make this assumption) in the joint-representation. Some representational gaps may be very large, for example, when there is a difference in the overall conceptualization of the problem (e.g., a sales problem vs. a cost reduction problem). A representational gap can also occur within a narrow range of the problem where two or more parties perceive a specific issue differently (e.g., the meaning of market data). No matter the size, a representational gap always occurs between the REPRESENTATIONAL GAPS AND CONFLICT 12 joint-representations of two or more people. The size would only reflect how many goals, assumptions, and operators are misaligned. Representational gaps are likely to arise when people have different ways of seeing the world; this is likely when there is diversity in individual teammates’ backgrounds (Cagan & Vogel, 2002; Weingart, Cronin, Houser, Cagan, & Vogel, in press). Note that by diversity, we refer specifically to differences in thought worlds (Dougherty, 1992). Thought worlds are how a person interprets their environment, including what one pays attention to, and what relationships one assumes to exist. Thought worlds are used to make sense of the environment and its operation, and are a product of one’s education and experience solving problems over one’s career. Dougherty (1992) and Weingart et al. (in press) have described how a person’s functional background (e.g. engineering, industrial design) defines a thought world, but we would also expand it to include other factors that shape people’s world views such as age group and culture (both organizational and national). For example, someone from IBM where the culture is well structured and conservative will approach software development very differently than someone from Microsoft, known for having a more free-flowing and open environment. Similarly, we would expect the Japanese to conceive of negotiations very differently than Americans (Gelfand et al., 2001). Different thought worlds lead people to construct different problem representations. Dearborn & Simon (1958) demonstrated this in their classic study of how executives saw organizational problems from the perspective of their own departments (e.g., finance managers saw a decrease in revenue as stemming from mismanagement of money, while marketing managers saw the same decrease in revenue as stemming from poor marketing of the product). Similarly, those from different thought worlds can consistently interpret a specific issue REPRESENTATIONAL GAPS AND CONFLICT 13 differently, again leading to different problem representations. For instance, bankers see risk as avoidable and uncontrollable and therefore make risk minimization an important goal; entrepreneurs see risk as unavoidable but controllable and therefore pay little attention to risk, focusing instead on controlling the market (Sarasvathy, 2001). The reason that people’s thought worlds evoke different representations in a given situation is that any particular discipline tends to focus on some segments of the task environment to the exclusion of others. The segments that attract attention are partially a function of the knowledge domain for the particular function. Designers learn about form but not physics, thus they pay attention to aesthetics of materials but not solid dynamics. The elements are therefore different (e.g., color and shape rather than weight and tensile strength), as are the operators used to change them. The value system of a thought world also directs where people focus their attention. Engineers earn rewards for making things work correctly, not for making them beautiful, so they pay attention to function and not form. This affects the goals they create (e.g., durability in use will be more important than shapeliness of the object) and the assumptions they make. Regardless of the specifics, the more functionally diverse a team and the less the functional backgrounds share knowledge and values, the higher the likelihood that the jointrepresentations of the people in the group will be nontrivially different, and there will be representational gaps. We also speculate that the nature of the gaps will depend on whether they reflect dissimilarities in the knowledge or in the values between thought worlds. Representational gaps based on different knowledge sets are likely to surface in omissions rather than direct conflicts. Because thought worlds tend to reflect knowledge domains that are closed systems, knowledge that is central to one is likely to be irrelevant to another. Engineers do not REPRESENTATIONAL GAPS AND CONFLICT 14 think about design concepts because they fall outside “engineering concerns”. In practice, this means that a teammate may bring the problem into what is an illegal (or unacceptable) state for another team member. For example, at GiPSE a designer placed the company’s logo, made from wood inlay, on the center of the steering wheel. When seeing this mock-up, an engineer quickly pointed out that deployment of the airbag would turn the logo into a projectile, possibly injuring the driver. The designer, whose thought world was less focused on physical safety, made no assumptions about what happens during a crash, nor did he have operators to apply to tell him what would happen in such a situation. The omission of this knowledge made the designer’s joint representation inconsistent with the engineer’s (who commonly deals with safety issues and who would make assumptions about what happens during a crash). In contrast, representational gaps that are based on different value systems are more likely to emerge as direct conflicts rather than mere omissions. The set of issues that are the subject of value judgments tends to be relatively constant across thought worlds (i.e., most functions have a weighting for the relative importance of cost, durability, styling, and innovativeness). These issues tend to be weighted differently by different thought worlds. Thus people who weigh an issue differently are likely to be in conflict over what to do. For example, a designer, concerned about the visual line of an interior door panel, became upset when an engineer placed a speaker in a location that interfered with the line. The engineer had placed the speaker there to optimize the layout of the components behind the door panel (valuing effective use of space over aesthetics) whereas the designer did not approve because of the impact on the visual line (valuing the aesthetic over the effective use of space). P1. Thought world diversity in a team will lead to representational gaps among team members. REPRESENTATIONAL GAPS AND CONFLICT 15 P1a. Representational gaps stemming from different knowledge bases across thought worlds will primarily exist as omissions (one side acknowledges an issue where the other side does not). P1b. Representational gaps stemming from different value systems across thought worlds will primarily exist as contradictions (both sides acknowledge an issue but disagree on its importance or relevance). In both of the examples above, the actual moves made by a team member (putting a wood logo on the steering wheel or moving the location of the speaker) made team members disagree as to whether the move was acceptable. This is how representational gaps lead to conflict. Although one could comment that such disagreements can also be opportunities to synthesize each other’s views into a creative solution (cf. Cramton & Hinds, in press), we will argue that the nature of representational gaps is such that they are more likely to lead to conflict than learning. REPRESENTATIONAL GAPS AND CONFLICT People with incompatible joint-representations are effectively solving incompatible problems; hence they are likely to have conflicting ideas about how to solve the team’s problems. In a situation where people are deciding together which move to select, they will have different valuations of the outcome of a move when the underlying goal hierarchies are different; or a move may seem implausible if one is unaware of the existence of the operators on which it relies. It should be no surprise, therefore, that one team member’s actions can frustrate the work of another. The actions can conflict directly – one team member might try to add features to the product to increase value while another tries to reduce features to minimize cost. The actions can also conflict more subtly through loosely coupled actions, for example if one ignores a deadline, it will result in another having to rush his or her job. P2. Representational gaps will increase conflict in teams. Conflict due to representational gaps need not only occur over the actions a team is to take. Because a representation is used to interpret information, representational gaps will lead to REPRESENTATIONAL GAPS AND CONFLICT 16 conflict over the meaning of new information. For instance, after a deadline for completing a component has been extended, a designer who does not believe in a linear schedule may interpret the extension as an endorsement to redesign the part to what the designer had envisioned. An engineer may interpret this same extension as a mandate to refine the current design in order to get back on schedule; this interpretation being in line with the engineer’s assumptions and goals. Not only will they interpret information differently, they may judge its value differently. For example, engineers and designers judge customer survey data very differently because they have different assumptions. Engineers, who assume the customer knows what they want, will take this information seriously; designers, who assume the designer will show customers what they should want, will dismiss it entirely. Thus representational gaps lead to disagreements over what teammates should conclude as well as what they should do. P2a. Representational gaps lead to conflict via differences in how teammates interpret information. Conflict resulting from representational gaps will, on the surface, resemble task conflict. People will disagree over the right way to progress, and will need to come to some satisfactory agreement to proceed. However, reducing conflict stemming from a representational gap is challenging because the source of the conflict is in the goals, assumptions, elements, and operators, and direct conflict over GAEOs is unlikely to surface. It is much more likely that conflict will repeatedly surface in disagreements over which moves to make; the particular disagreement being a consequence of the underlying representational gap. When there is a disagreement over how to proceed, people are less likely to question whether others share their fundamental beliefs about the world because they have a tendency to believe that others think as they do (Krueger & Clement, 1994; Ross, Greene, & House, 1977). In other words, it may not occur to an engineer to ask whether a designer shares his or her belief that functionality is the REPRESENTATIONAL GAPS AND CONFLICT 17 primary predictor of customer satisfaction. Instead, design will continue to make moves that favor styling as the primary goal while engineering makes moves that favor functionality. P2b. Conflict stemming from representational gaps is more likely to focus on the current moves than the underlying goals, assumptions, operators, and elements. When conflict arises, it can baffle team members as to why others do not see the logic of their choices. In addition to teammates assuming others see the world as they do, the “teamwork” context may cue team members to expect their joint-representations to be the same. Unlike a negotiation, where the context cues people that each side possesses different goals and restrictions (Pinkley, 1990), people working together on the same team will have an expectation that they should see things similarly because they are supposed to be working under the same allegiance and solving the same problem. Moreover, since individuals tend not to share their unique information, especially in teams solving problems with no right answer (Wittenbaum & Stasser, 1996), representational gaps are unlikely to surface. We expect that it will take time and repeated disagreements before parties realize their conflict stems from having different goals or assumptions. In the beginning, the move conflict may just be considered a normal part of group interaction. Over time, frustration can increase because people will sense that they have “been here before” and will be tired of arguing over a particular issue. This frustration may provoke people to look deeper and uncover the representational gap at the root of the problem, but not before time is wasted and frustration rises. Worse, people may make negative attributions about why their teammates are being difficult, focusing on personality rather than situation (Kelley, 1971). The implication is that even though a disagreement is an opportunity to integrate differing perspectives, by the time people have discovered the representational gap, they are more positional and defensive than open and accommodating. REPRESENTATIONAL GAPS AND CONFLICT 18 Reducing Representational Gap Conflict On teams, where people are working toward cooperative goals and may be in ongoing relationships, integrative bargaining techniques are optimal for managing conflict (Fisher, Ury, & Patton, 1991; Weingart & Jehn, 2000). In particular, focusing on parties’ underlying interests will be especially helpful in crafting agreements given the complex problems that multifunctional teams face (Weingart et al., in press). However, integrative bargaining will prove to be tricky in teams with representational gaps because the different thought worlds among people in the diverse teams reduce both the ability and the motivation to use the information others share. P3. Representational gap-based conflict is more difficult to bridge than other forms of task conflict. For one person to successfully incorporate the other’s interests into an agreement, that person must have some understanding of what satisfying the other’s interest implies, and some ability to gauge the tradeoffs that must be made in order to accommodate the interest. Because thought worlds are somewhat self-contained, the interests of one may not be easily connected to those of another. For instance, an engineer may hear that “maintaining the vehicle theme is our most important priority” from the designers, but what “maintaining the vehicle theme” means, exactly, is not obvious. Themes are not in the lexicon of engineering, and so what this implies for the engineer is not clear. Further, even if the engineer understands that “theme” means that components share similar shapes, lines, colors, and materials, the engineer (who does not deal in those terms) will have almost no ability to know how to trade these off against weight, cost, durability, and ease of manufacturing. The same problem of understanding holds true when trying to use standards to resolve differences. The knowledge about what and how a standard applies is unlikely to cross thought world boundaries. REPRESENTATIONAL GAPS AND CONFLICT 19 P3a. Using interests and standards to resolve task conflict will be only minimally effective if the conflict stems from representational gaps. One may argue that teammates could use matters that are relevant to multiple thought worlds to help gauge tradeoffs, for example if the ultimate aim of GiPSE is to delight the customer, people could make their tradeoffs based on what is more delightful to the customer. Carlile (Carlile, 2002; 2004) has addressed this idea formally with the notion of boundary objects: concepts or attributes common to multiple thought worlds that can be used as a means for translation and calibration for matters unique to each thought world. At GiPSE, the customer is a boundary object. The customer’s interests could potentially be used to scale the importance of any function’s interests. Unfortunately, boundary objects are still interpreted through the lens of a thought world, so the disagreement based in the representational gap is merely relocated from the boundary object itself to the implications of that object. In our examples, factors influencing customer’s delight becomes interpreted through a function’s belief system (e.g., designers may recognize that vehicle functionality affects purchase decisions, but they will simply degrade the magnitude of this effect while bolstering the importance of aesthetics, thus interpreting the customer’s perspective so that it matches their own values). P3b. Interests and standards, even when relevant to multiple thought worlds, will be interpreted so that they align with the values of a thought world and hence will reflect the representational gap. In addition to developing a shared understanding and therefore providing team members with the ability to integrate other’s interests and standards to arrive at an agreement, there is also the issue of motivating people to integrate what others communicate. As Tetlock (2000, p. 317) argued, “[understanding others’ positions]... should not be confused with tolerance, less still willingness to change one’s mind.” Team members are likely to believe others should think as they do. Part of this is due to context; teamwork is unlike a negotiation between separate parties REPRESENTATIONAL GAPS AND CONFLICT 20 in that people do not expect their teammates, by definition, to take an opposing view. Another part of this belief is that people are able to see many reasons in support of one’s own position and few if any in support of the other side’s position. People’s own ideas are connected to a large body of existing knowledge (Heath, Larrick, & Klayman, 1998), whereas knowledge from outside a thought world is not. Because of this, people may reject other’s goals and assumptions because they fail to understand how these will help the problem at hand. People can also reject other’s views based on less logical criteria. Tetlock (2000) showed that manager’s cognitive style (conceptualized as one’s preference for simple unambiguous answers as opposed to complex ambiguous ones) affected the way they “spun” a situation in order to support their own conclusion. These cognitive styles reflected different value sets. In the same way, we expect thought world derived values to bias the way people interpret and evaluate a situation. Although rejecting another’s view based on such a value laden analysis should not be irrational, it is hardly objective, as the basis for judgment is an unshared value system. Alternately, people may feel that their own perspective is superior to the other person’s just because it is their own. In part, this is attributable to an egocentric bias (Krueger et al., 1994; Kruger, 1999). In general, however, people tend to weight the opinions of others less heavily than their own (Yaniv, 2004), and this tendency increases the more discrepant that opinion is from one’s own (Yaniv & Kleinberger, 2000). This preference can be thought of as essentially “my viewpoint is better because it is mine”. Although little research has been done to examine the application of this principle to ideas, one can draw parallels to people’s non-rational preference for objects they possess (the endowment effect, see (Franciosi, Kujal, Michelitsch, Smith, & Deng, 1996), and groups to which they belong (Tajfel, 1974). P3c. Information sharing will only be minimally effective in reducing representational gaps because of a lack of shared understanding. REPRESENTATIONAL GAPS AND CONFLICT 21 P3d. Information sharing will only be minimally effective in reducing representational gaps because of a lack of motivation to use the information. The Moderating Effects of Cognitive and Affective Integration The recurrent and resilient nature of representational gap conflict is frustrating to those involved, especially when the disagreements challenge the legitimacy of people’s worldviews. Thus, representational gaps increase the chance that task conflict will transform into relationship, or interpersonal conflict and potentially derail problem solving (Simons & Peterson, 2000). With this in mind, we propose two methods to reduce representational gap conflict, cognitive and affective integration. We briefly review each here to encourage research in these areas. Cognitive integration. Two or more people are cognitively integrated the more that one is capable of analyzing a situation using another’s thought world. They do not have to share a thought world; rather a person must understand another’s thought world. That is, one can connect another’s jargon, knowledge, and values to those of his or her own thought world. Three things are important to note regarding cognitive integration as a construct. First, whereas representational gaps refer to differences in problem representations, cognitive integration refers to understanding of others’ thought worlds. Thought worlds exist independent of the task at hand, whereas representational gaps do not. Second, cognitive integration is not the opposite of thought world diversity. Teams with the same level of thought world diversity can vary in their level of team member understanding of one another’s thought worlds, and thus experience different levels of cognitive integration. Third, training a person to understand another’s thought world does not mean that they will reach the same conclusions, it just means that they will be capable of seeing the structure of the problem similarly due to their increased understanding. Some organizational functions are naturally more cognitively integrated than others. Functions such as engineering and finance share an epistemic style (e.g., a belief in the numbers REPRESENTATIONAL GAPS AND CONFLICT 22 and problem decomposability), and would have more overlap in their thought worlds than, say, design and finance (design having neither belief). Group members who frame problems differently due to their culture, training, or education can become cognitively integrated by learning about others’ perspectives either through experience or training. Through this understanding, an individual can configure the GAEO of another person into his or her own representation of the problem, effectively reducing the likelihood that the existing thought world diversity will result in representational gap. For example, designers who are cognitively integrated with engineers will anticipate that using metal for a component, although more stylish, will have consequences on the weight of the component. Thus, people may still disagree over particular moves (whether or not to use metal), but these disagreements will be more easily resolvable through an appeal to common assumptions and goals. Cognitive integration implies that diversity will be less likely to result in representational gaps, as people will seek to be mindful of what others on the team will need. Thus, there will be fewer conflicts. In addition, when there are conflicts, they are more likely to be traditional task conflict – conflict over moves – which can then be resolved through integrative bargaining (Fisher et al., 1991). That is, fewer representational gaps imply more shared GAEO, and these can be used to understand and integrate ideas between people, much in the way an individual’s problem representation makes adaptation of data to a given problem possible. P4. Cognitive integration reduces the likelihood that thought world diversity will lead to representational gaps. P4a. Cognitive integration increases the level of shared understanding in a team via the ability to connect another’s jargon, knowledge, and values to those of one’s own thought world. One should recognize that there can be costs to cognitive integration. Some of these costs come from front-end training. It takes time to train a team member to understand the concerns of REPRESENTATIONAL GAPS AND CONFLICT 23 another. Some may even argue that doing so dilutes team members’ expertise. For example, when a designer becomes overly concerned with timelines, he or she might lose sight of design goals and the vehicle’s appearance might suffer. Team members serve an advocacy role in addition to bringing their functional expertise to the table and so it would be a legitimate concern that cognitive integration may decrease the attention devoted to one’s own functional area interests. Finally, when there is too much cognitive integration a team risks reducing the potential for creative solutions because as team member’s representations become more similar, it reduces the likelihood that one will encounter unique knowledge when interacting with others. Cognitive integration reduces conflict by increasing shared understanding and thereby reducing the impact of thought world diversity on representational gaps. However, some representational gaps may be irreducible. This can be because teams want to use different perspectives to maintain constructive controversy (Tjosvold, 1985). People may also lack the time or training to fully understand other’s thought worlds. Alternately, people can choose to maintain their view of the problem and see the other perspective as incorrect. In these cases the problem is not understanding but motivation. Encouraging motivation requires affective integration. Affective integration. Affective integration relates to team members’ interpersonal relationships, in particular the levels of liking, respect, and trust amongst team members. Liking is a general positive feeling that attracts one person to another (Chaiken, 1987). If I like someone, I would choose to spend time with him or her, and would enjoy his or her company. Respect is esteem for a person based on his or her abilities and character (Cronin, 2004). If I respect someone, I believe that he or she deserves attention and consideration, even if I disagree with him or her. Trust is the belief that the other person tells the truth (Hass, 1981), and will not REPRESENTATIONAL GAPS AND CONFLICT 24 harm me even if I cannot monitor their behavior (Mayer, Davis, & Schoorman, 1995). If I trust someone, I will be vulnerable to him or her because I believe there is no malice or duplicity in his or her motives. The more people trust, respect, and like each other, the more they are affectively integrated. Affective integration is not about thinking style or domain knowledge, it is about how people think and feel about each other. Affective integration develops over time as people’s relationships develop. Affective integration uses people’s attitudes toward one another to provide motivation for accepting what the others say or want. It does so by providing the basis for inferences about veracity and value of the other party’s information. Trust can imply that other team members’ beliefs are true (Hass, 1981) and not malicious (Mayer et al., 1995). Respect can imply that other team member’s views have value and deserve consideration (Cronin, 2004). Liking may also have a weak effect on acceptance as people become more positive toward familiar arguments (see Petty & Wegener, 1998 for a review). The positive evaluation with which affective integration imbues others’ ideas can lead people to yield to others when they disagree. An ancillary benefit is that people are less likely to get mired in trying to convince one another of their respective positions, which has been shown to have a negative effect on conflict resolution (Hyder, Prietula, & Weingart, 2000). Instead, one recognizes that another’s belief is different, and that he or she may not understand it, but they are content to defer to the wishes of the other. When people are not affectively integrated, they will dismiss others’ ideas and potentially judge them as false or worthless. For example, often engineers do not want to alter their specifications to accommodate designers because they do not trust that the aesthetic changes matter to the designers as much as designers say. “Every change REPRESENTATIONAL GAPS AND CONFLICT 25 is ‘absolutely necessary’,” quips one engineer, explaining why he does not try to reengineer a component to accommodate the design change. P5. Affective integration reduces the effect of representational gaps on conflict. P5a. Affective integration increases the value team members place on others’ ideas and motivates them to accept those ideas. The fact that affective integration uses an inferential process rather than a data driven one to evaluate an idea means that affective integration should be very useful in getting people to choose to accept other’s ideas. As argued earlier, there can be many issues that a particular thought world is not prepared to understand (e.g., engineering has no framework for “vehicle theme”) and therefore may be rejected because they do not fit with an individuals own thought world. Getting a person to accept these ideas will be more a matter of faith (belief in the absence of proof) rather than data. Affective integration provides such faith. Affective integration should also help ameliorate the “my argument is better because it is mine” effect by giving people a reason to concede their own beliefs to others. One who is affectively integrated with another will see their relationship as providing a reason to yield to that person’s ideas. An implication of this argument is that affective integration can also indirectly reduce representational gap conflict by motivating people to understand each other’s perspective (i.e., improve cognitive integration). Understanding takes work (e.g., a designer would not automatically comprehend engineering principles), and so a person must be motivated to put the effort into understanding another’s perspective. Liking can help in this capacity, as it will provide a reason for people to spend time together, in turn leading to more opportunities to share information. In the contrary case, if people do not like each other, they would be more likely to withdraw. In addition, people should believe that there would be a benefit to learning the perspective of another. In this regard, respect will help because respect implies that one believes REPRESENTATIONAL GAPS AND CONFLICT 26 there is value in what the other person says, and one will be more inclined to try to find that value. Trust should also affect understanding, as if one does not believe what the other says, one may try to impute the “real” meaning or may dismiss it as false. P6. Affective integration increases the level of cognitive integration of the team by motivating teammates to try to understand others’ perspectives. DISCUSSION In this paper, we describe the phenomena of representational gaps; their source, their consequences for group performance, and ways to ameliorate their negative effects. The notion of representational gaps may partially explain the mixed support for the benefits of diversity (Milliken et al., 1996; Williams et al., 1998) as they can explain why some diverse groups are unable to capitalize on the disparate knowledge of a diverse team. Representational gaps may also contribute to explain why task conflict is most often negative (De Dreu et al., 2003b) – representational gaps decrease the value of information sharing. Because representational gaps occur in the a priori definition of a problem and reflect conflicting fundamental beliefs between those who have the gap, representational gaps are hard to ameliorate. We have therefore proposed two methods of team integration, cognitive and affective, that we believe should help in this regard. By understanding the dynamics in our proposed model (Figure 1) we expect not only to improve how diverse teams work, but teams and collaborative work in general, as many of the issues we have raised can be applied in other contexts as well. Addressing Representational Gaps The first step in addressing representational gaps in a team is finding them. As we have argued, representational gaps may not surface until conflict becomes protracted, during which time the team’s work, relationships, and members’ personal well-being may suffer (De Dreu, Van Dierendonck, & De Best-Waldhober, 2002; De Dreu, Van Dierendonck, & Dijkstra, 2004; REPRESENTATIONAL GAPS AND CONFLICT 27 De Dreu & Weingart, 2003a). Yet it would be easy to identify teams with a high likelihood of representational gaps: those with members of very different thought worlds. Simple awareness that others may not look at the same elements, make the same assumptions, or have the same goals may be helpful in promoting a search for representational gaps. The next step would be to determine exactly where those gaps occur. Team members might be able to gain this insight by keeping a journal of disagreements and looking for patterns over time. Journal entries could include the content of the disagreement (or what triggered the conflict), information on who was involved, when it occurred relative to external deadlines, and why it happened. Over time the team should attempt to determine whether conflict included the same person or function or was over a certain type of issue, which might imply an underlying representational gap. Team members then need to question their attributions for the conflict, preferably giving each other the “benefit of the doubt” as to whether the conflict is merely argumentative. This will encourage teammates to work together to determine whether the conflict is attributable to different problem definitions. Once a representational gap is identified, it will be important to increase cognitive and affective integration. Affective integration requires the development of trust, respect, and liking. These may develop naturally over time. Yet if they do not or there are problems in the group, it will be crucial to find ways to facilitate their development. Some activities already exist; there are a myriad of exercises designed to develop trust within teams (e.g., Newstrom & Scannell, 1980). The large majority of these activities involve exercises outside of the work context and research is needed to determine their impact on affective integration in the workplace. Affective integration might also be increased by promoting activities to help develop liking via personal and familial bonds in a non-work context, e.g., social gatherings for team members and their REPRESENTATIONAL GAPS AND CONFLICT 28 families. Again, there are a number of research avenues to explore in terms of whether these activities actually develop trust, respect, or liking, as well as the mechanisms through which they operate. For instance, is familiarity enough to maintain liking in the face of an ongoing relationship? Can one have trust without respect? Which of these develop first? Are the development processes asymmetrical (e.g., easier to lose than gain)? In addition, there may be other ways to develop personal relationships that affect trust respect, and liking indirectly. For example, increasing cohesion may lead people to opportunities to increase trust, respect, and liking. Alternately, it may be possible to reduce the salience of the subgroups (or faultlines) within the team (Cramton et al., in press; Lau & Murnighan, 1998). It may be possible to make other distinctions more salient (such as team membership, a shared superordinate goal, or other characteristics that are shared by team members but differentiate themselves from others) so that people feel less split into factions or more competitive with people outside the group. Future research could address how to minimize the salience of some characteristics while maximizing the salience of others. Alternately, Dougherty & Takacs (2004) describe “team play”, a process by which groups engage in experimentation, heedful interrelating and ultimately innovation. Such heedful interrelating requires a combination of trust, respect, and liking. Future research could look at the various ways each of the subcomponents of affective integration relates to team play. As people interact and if affective integration increases, cognitive integration is likely to increase as well. If not, cognitive integration might be improved by having one team member shadow another while the target person goes through his/her daily routine. Job shadowing can provide insight into the types of demands and decisions teammates face on a regular basis. The target person should explain his/her perspective on each given situation when transitioning from REPRESENTATIONAL GAPS AND CONFLICT 29 one activity to the next (a talk-aloud type of technique). Alternately, the shadowing person can ask questions when puzzled by the teammates activities, which promotes the practice of asking for clarification in times of uncertainty. Team members can interview one another to try to understand what goals, constraints, or challenges others face in performing their job. This will provide insight into how a teammate’s environment shapes the way they think about the problems they face. When team members provide one another with information about the methods they use to decompose and solve problems, it can demonstrate the underlying assumptions that drive their analytical approach. The important question, in addition to how to increase cognitive integration, is how to find the optimal level of cognitive integration. One should recognize that there are costs associated with cognitive integration. Creative solutions require incorporating new knowledge that changes the problem representation (Ohlsson, 1992). When team members begin with similar representations, it reduces the likelihood that one will encounter unique knowledge. Yet not all tasks have the same potential use for new and creative ideas. For example, discovering a better way to run an assembly line could be wasted if the assembly line has already been engineered. Alternately, when looking for new technological applications, breadth of thought is of critical importance. The optimal amount of cognitive integration will likely depend on certain facets of the task that the team is to perform, and these have yet to be identified. Generalization Finding the representational gaps, improving affective integration and finding the optimal level of cognitive integration should help multi-functional teams take advantage of their diversity while at the same time making them less stressful to be in. Yet the issues discussed in this paper are generalizable beyond multi-functional teams. Understanding how these issues operate will REPRESENTATIONAL GAPS AND CONFLICT 30 help other types of teams and contribute to more general theory on cognition, group dynamics, and conflict management The issue of conflict stemming from different thought worlds is not exclusively the province of multi-functional teams. Although we have concentrated on multi-functional teams, the theory is applicable to any situation where people with different knowledge bases come together to solve a problem. All kinds of diversity can lead people to have different thought worlds. A large body of literature on demographic diversity supports this conclusion (Williams & O’Reilly, 1998). Cognitive style differences (Tetlock, 2000) could also be seen to represent different thought worlds. Cramton (2001) has described how even geographic locations can beget different thought worlds, as they do not draw from a large base of common knowledge and may thus take assumptions they make about norms (e.g., people take public transportation) or even day to day functioning (e.g., it was snowing in Finland so team members were late to work) for granted. Note also that representational gaps may occur in what appear to be homogenous teams. People often have different fundamental beliefs because they have different personalities and life experiences. The effect of an interpersonal representational gap will likely be similar to social category-based ones. It would be instructive to see how the management of interpersonal based representational gaps is different from ones that occur between people from different social categories because people within similar social categories have more reason to believe they should think similarly, they may be even harder to recognize. The role of affective and cognitive integration should help researchers understand healthy team functioning beyond conflict. A team that has both cognitive and affective integration is likely to have what Cramton & Hinds (in press) call mutual positive distinctiveness – “the extent REPRESENTATIONAL GAPS AND CONFLICT 31 to which the group respects differences among members in views, values, competencies, and practices and sees these differences as a potential source of advantage for the group as a whole” (p. 23). As Cramton and Hinds (in press) have argued, diverse groups with mutual positive distinctiveness will be less likely to splinter into ethnocentric subgroups and more likely to engage in cross-group learning. Cramton and Hinds (in press) define diversity broadly, and under their conceptualization mutual positive distinctiveness would apply to teams where members have salient differences between member’s characteristics, which is likely to be the majority of teams. This argues both for the value in understanding mutual positive distinctiveness, as well as how it can be achieved through cognitive and affective integration. The interface between affective integration and cognition is another potentially fruitful area of research, as these have been many links between affect and thinking (Forgas, 1995) as well as organizational behavior in general (Brief & Weiss, 2002). A lack of cognitive integration may influence affective integration; we may have less trust for another who uses a different language and has different priorities. At the same time, affective integration may affect people’s thinking in and of itself. Cronin (2004) showed that a lack of trust or respect for a source led a person to think negative thoughts in response to that source’s ideas. In a sense, this would apply the persuasion paradigm (see Eagly & Chaiken, 1998; Petty et al., 1998) to show how “objective” thought (i.e., what to do to solve a problem) is affected by the attitudes people hold. Understanding how people alter their existing representations to accommodate new information is understudied (Simon, 1991). Yet that process is the same as that of closing a representational gap. When people revise their framework for solving a problem, it represents a type of double loop learning, i.e., questioning underlying values and assumptions (Argyris & Schon, 1978). As such there are likely to be the issues of defensiveness that Argyris and Schoen REPRESENTATIONAL GAPS AND CONFLICT 32 identify with regard to double loop learning. In addition, detecting a representational gap may also have more features of double loop than single loop learning in terms of the underlying cognition. In conclusion, by integrating theories of information processing, conflict management, and group dynamics, we can gain increased insight into a root cause of conflict in diverse teams. We believe representational gaps are prevalent in diverse teams. Identifying and ameliorating these gaps should allow groups to more effectively tap into their diversity and improve their functioning. REPRESENTATIONAL GAPS AND CONFLICT 33 REFERENCESAmabile, T. M. 1988. A model of creativity and innovation in organizations. Research inOrganizational Behavior, 10: 123 167.Anderson, J. R. 1993. Rules of the Mind. Hillsdale, NJ: Lawrence Erlbaum.Argote, L., Gruenfeld, D. H., & Naquin, C. 2001. Group learning in organizations. In M. E.Turner (Ed.), Groups at Work: Advances in Theory and Research: 369-411. Mahwah,NJ: Lawrence Erlbaum.Argyris, C., & Schon, D. 1978. Organizational Learning: A Theory of Action Perspective.Reading, MA: Addison-Wesley.Brief, A. P., & Weiss, H. M. 2002. Organizational behavior: Affect in the workplace. AnnualReview of Psychology, 53: 279-308.Cagan, J., & Vogel, C. M. 2002. Creating Breakthrough Products. Upper Saddle River, NJ:Prentice-Hall.Cannon-Bowers, J. E., Salas, E., & Converse, S. 1993. Shared mental models in expert teamdecision-making. In J. Castellan (Ed.), Individual and Group Decision Making: CurrentIssues: 221-247. New York: LEA Pubishers.Carlile, P. R. 2002. A Pragmatic View of Knowledge and Boundaries: Boundary Objects in New Product Development. Organization Science, 13: 442-455. Carlile, P. R. 2004. Transferring, Translating, and Transforming: An Integrative Framework forManaging Knowledge Across Boundaries. Organization Science, 15: 555-568.Chaiken, S. 1987. The heuristic model of persuasion. In M. P. Zanna, & J. M. Olson (Eds.), Social influence: The Ontario symposium, Vol. 5: 3-39. Hillsdale, NJ: LawrenceErlbaum Assoc., Inc. REPRESENTATIONAL GAPS AND CONFLICT 34 Chi, M. T. H., Feltovich, P. J., & Glaser, R. 1981. Categorization and representation of physicsproblems by experts and novices. Cognitive Science, 5: 121-152.Cramton, C. D., & Hinds, P. J. in press. Subgroup Dynamics in Internationally DistributedTeams: Ethnocentrism or Cross-National Learning? In S. B.M., & R. Kramer (Eds.),Research in Organizational Behavior.Cronin, M. A. 2004. The effect of respect on interdependent work. Unpublished DoctoralDissertation, Carnegie Mellon University.De Dreu, C. K. W., Van Dierendonck, D., & De Best-Waldhober, M. 2002. Conflict at work andindividual wellbeing. In M. Schabracq, J. A. M. Winnubst, & C. L. Cooper (Eds.),International Handbook of Work and Health Psychology: 495-515. Chichester, UK:Wiley.De Dreu, C. K. W., Van Dierendonck, D., & Dijkstra, M. T. 2004. Looking back, looking ahead:Conflict at work and individual health and well-being. International Journal of ConflictManagement, 15: 1-18.De Dreu, C. K. W., & Weingart, L. R. 2003a. A Contingency Theory of Task Conflict andPerformance in Groups and Organizational Teams. In M. West, D. Tjosvold, & K. Smith(Eds.), International Handbook of Organizational Teamwork and Cooperative Working: 151-165. Chichester, England: Wiley. De Dreu, C. K. W., & Weingart, L. R. 2003b. Task Versus Relationship Conflict, TeamPerformance, and Team Member Satisfaction: A Meta-Analysis. Journal of AppliedPsychology, 88: 741-749. Dearborn, D. C., & Simon, H. A. 1958. Selective perception: A note on the departmentalidentification of executives. Sociometry, 21: 140 144. REPRESENTATIONAL GAPS AND CONFLICT 35 Dougherty, D. 1992. Interpretive barriers to successful product innovation in large firms.Organization Science, 3: 179-202.Dougherty, D., & Takacs, C. H. 2004. Team play: Heedful interrelating as the boundary forinnovation. Working Paper, Rutgers University.Eagly, A. H., & Chaiken, S. 1998. Attitude structure and function. In D. T. Gilbert, & S. T. Fiske(Eds.), The handbook of social psychology, 4th ed., Vol. 1: 269-322. New York:McGraw-Hill.Espinosa, J. E., & Carley, K. M. 2001. Measuring team mental models. Paper presented at the AOM, August 2001 in Washington, DC.Fiol, M. C. 1995. Thought worlds colliding: the role of contradiction in corporate innovationprocesses. Entrepeneurship: Theory and Practice, 19: 71-91.Fisher, W., Ury, W., & Patton, B. 1991. Getting to Yes. New York: Penguin Books.Forgas, J. P. 1995. Mood and judgment: The affect infusion model (AIM). PsychologicalBulletin, 117: 39-66.Franciosi, R., Kujal, P., Michelitsch, R., Smith, V., & Deng, G. 1996. Experimental tests of theendowment effect. Journal of Economic Behavior & Organization, 30: 213-227.Galbraith, J. 1977. Organization Design. Reading, MA: Addison-Wesley. Gelfand, M. J., Nishii, L. H., Holcombe, K. M., Dyer, N., Ohbuchi, K.-I., & Fukuno, M. 2001. Cultural influences on cognitive representations of conflict: Interpretations of conflictepisodes in the United States and Japan. Journal of Applied Psychology, 86: 1059-1074.Hass, R. G. 1981. Effects of source characteristics on cognitive responses in persuasion. In R. E. Petty, T. M. Ostrom, & T. C. Brock (Eds.), Cognitive responses in persuasion: pp. 141-172. Hillsdale, NJ: Erlbaum. REPRESENTATIONAL GAPS AND CONFLICT 36 Hayes, J. R., & Simon, H. A. 1974. Understanding written problem instructions. In L. W. Gregg(Ed.), Knowledge and Cognition: 167-200. Hillsdale, NJ: Lawrence Erlbaum.Heath, C., Larrick, R. P., & Klayman, J. 1998. Cognitive repairs: How organizational practicescan compensate for individual shortcomings. Research in Organizational Behavior, 20:1-37.Hutchins, E. 1990. The technology of team navigation. In J. Galegher, R. E. Kraut, & C. Egido(Eds.), Intellectual Teamwork: Social and Technological Foundations of CooperativeWork: 191-220. Hillsdale, NJ: Lawrence Erlbaum Assoc., Inc.Hyder, E. B., Prietula, M. J., & Weingart, L. R. 2000. Getting to Best: Efficiency versusOptimality in Negotiation. Cognitive Science, 24: 169-204.Kelley, H. H. 1971. Attributions in social interactions. Morristown, NJ: General Learning Press.Klimoski, R., & Mohammed, S. 1994. Team mental model: Construct or metaphor. Journal ofManagement, 20: 403 – 437.Krueger, J., & Clement, R. W. 1994. The truly false consensus effect: An ineradicable andegocentric bias in social perception. Journal of Personality & Social Psychology, 67:596-610.Kruger, J. 1999. Lake Wobegon be gone! The below-average effect and the egocentric nature of comparative ability judgments. Journal of Personality & Social Psychology, 77: 221232.Lau, D. C., & Murnighan, J. K. 1998. Demographic Diversity and Faultlines: The CompositionalDynamics of Organizational Groups. The Academy of Management Review, 23(3): 325-