CSR Processes in Governance Systems and Structures: The Development of Mental Modes of CSR

When corporate social responsibility (CSR) as a sensemaking process is assessed from a corporate governance perspective, this implies that stakeholders do not only influence companies by promoting and enforcing regulations and other corporate guidelines. They also influence companies by promoting regulation on influence pathways, by demanding that companies develop formal mechanisms that allow companies and stakeholders to discuss and in some cases agree on changes to principles and policies. This perspective suggests that regulation is an outcome of power relations and is, as such, a reflection of certain mental models. As such, mental models reveal the political bias in corporate governance perspectives. For this reason, CSR research needs to be clear about the underlying assumptions about corporate governance, and corporate governance research needs to disclose which mental models of CSR influence the outcomes. Taking a governance perspective on the development of mental models of CSR helps to understand the interaction between CSR and processes of sensemaking at the institutional, organizational and individual levels.


| 433 de GRAAF
The next section summarizes the emerging governance perspective on CSR in sensemaking processes. I illustrate these processes by drawing on some examples from the Netherlands. In Section 3, I describe the consequences of taking a governance perspective on CSR, and suggest that influence pathways and managerial discretion play critical roles in value attunement processes. I develop three propositions on CSR as a process of sensemaking within the interaction between governance systems, governance structures and individuals. The article concludes with a discussion of some of the implications of this emerging perspective.

MENTAL MODELS
Within this paragraph, the key definitions are defined and related to each other. Critical in my reasoning is the role of mental models. The human mind organizes and orders its experiences based on earlier experiences. These earlier experiences profoundly interact with current experiences, which leads to a "new" model of reality. As such, a mental model is socially constructing, and is constructed (Werhane, 2008;Werhane, Hartman, Archer, Englehardt, & Pritchard, 2013). These sensemaking processes (Weick, 1995) happen within governance systems and governance structures.

| Corporate governance and institutional conditions
If we assume that institutional conditions shape CSR (Aguilera et al., 2007;Campbell, 2007;Matten & Moon, 2008), then corporate governance mechanisms play a role in how those institutional conditions are expressed. The interaction between a company and its stakeholders is structured by a governance system; that is, the legal context in which a company's governance structure is embedded (Collier, 2007;de Graaf & Stoelhorst, 2013;Weimer & Pape, 1999). This governance system goes beyond corporate law. Governance systems include the legislation that has been developed to protect stakeholders, such as legislation on consumers' rights (Aguilera et al., 2007;Collier, 2007). Laws on corporate governance require companies to allow shareholders and, in a stakeholder-oriented system, other stakeholders, to influence their governance structure. As a result, influence pathways such as shareholder meetings are a legal requirement.
The governance structure of a company consists of all of the formal procedures and processes by which decisions are made (Nooteboom, 1999;Williamson, 2000). The literature on corporate governance emphasizes the role played by shareholders (Bezemer, Zajac, Naumovska, Bosch, & Volberda, 2015;Veldman & Willmott, 2015). This literature is consistent with and influenced by agency theory, which focuses on the power disparities between principals (shareholders) and agents (managers). Whereas shareholders have power as the owners of the capital, managers have more power owing to their position: they have more information about the company and the sector.
In most jurisdictions, the principal-agent dichotomy is not sufficiently complex to capture the legal context in which companies operate (e.g., Bezemer et al., 2015;Weimer & Paape, 1999;Whitley, 1999). In Northern Europe, for example, a company's works council has a say in corporate policies. Likewise, the agency perspective fails to take account of some organizational forms. Within a cooperative, for example, clients have a critical say. They could also be seen as shareholders, but in some jurisdictions, cooperatives take the form of associations that have no formal owner. In the Netherlands, for example, shares in a cooperative are held by a "dead hand": owned by no one. If such an organization is dismantled, the assets can be claimed by the state, not by the members of the cooperative (Galle, 1994).
In addition to statutory arrangements, a wide range of regulations shapes corporate principles and policies. This results in a set of complex institutional arrangements, some of which are explicit and binding, others implicit and cultural. Shareholders constitute one of the key groups. The number of groups that is represented depends on the institutional environment. With regard to certain issues, a Dutch works council has the right to ask for information (on almost everything), the right to be informed (on general strategy), the right to discuss corporate policies (on strategic issues) and a co-determination right (in the case of redundancies). The right to information alone can transform the relationship between employees and managers. Managers have to allow for the fact that they must inform employees in good time about important decisions, meaning that they tend to take the employees' perspective into consideration earlier than in shareholder-oriented systems (de Graaf & Herkströter, 2007). Comparable practices exist in other Northern European countries. Besides employees, governments and banks can also play an important role in the governance of a company. This is the case in China and Japan, for example (Weimer & Pape, 1999).
Aside from compulsory pathways of stakeholder influence, some companies also organize stakeholder influence pathways on a voluntary basis. In some countries, for example, executive boards have advisory committees on which important stakeholders are represented. Although they do not play a formal role, regular meetings ensure company-stakeholder interaction (de Graaf, 2016;de Graaf & Stoelhorst, 2013).
Besides corporate governance regulation, there is a wide range of other regulations that influence corporate policies. In addition to regulations on their employees and customers, companies have to follow regulations on the natural environment, trading conditions and production standards. From an institutional perspective, this institutional framework is an important determinant of CSR. Viewed from this perspective, CSR is no longer an "add-on," something that exists alongside a firm's economic and legal obligations, but the outcome of complex processes of responsiveness (Wood, 1991(Wood, , 2010. According to Wood, processes of responsiveness consist of stakeholder management, environmental analyses, and issue management. Taking an institutional perspective, we can add governance measures to these processes. Does this addition make stakeholder interaction processes unnecessarily complex, or does it allow us to further our understanding of CSR as sensemaking? I defend the latter view.

| Sensemaking, institutions and mental models of CSR
Sensemaking is "a retrospective process, driven by plausibility, social, ongoing, grounded in identity construction, enactive of sensible environments and focused on and by extracted cues" (Weick, 1995, p. 17). In this process, individuals deal with ambiguity and complexity, they deal with various interpretations of events, and they construct understandings of what is effective and ineffective, what is "good" and "bad," what should have been done, and how certain things should be done differently in future. When making important strategic, tactical, or operational decisions, for instance, a company's objective may be at stake, although this point may not be discussed officially with shareholders and other stakeholders. For example, if the company in question changes its brand strategy and wants to become known for its low prices rather than quality, at a certain point this will have implications for its suppliers. Managers have to decide which direction would best serve the company's goal, whatever that might be. This is a "searching process," for which there are no strict economic or legal guidelines. There is no one recipe for generating profits or serving stakeholder interests. Most often, this leads | 435 de GRAAF to a variety of objectives, in which profit plays a critical role, but that must also reflect the interests of customers and employees. This is an assumption of both shareholder perspectives (Jensen, 2001) and stakeholder perspectives on the nature of the firm (Freeman & Reed, 1983), two dominant mental models of the objective of the firm.
Whereas institutional theorists emphasize the stability of societies and structures, sensemaking is about processes, dynamics and change. Sensemaking is described as a process that lacks a clear beginning and end. Rooted within pragmatism, sensemaking theory assumes that thinking and action are interrelated, not only for an individual, but also within the social environment in which the individual is active (Weber & Glynn, 2006). In this respect, institutions seem to function like the bed of a river in which there is a constant flow of thoughts (opinions) and actions (e.g., Weick, 2012).
Within this constant dynamic, there is a need for certain "givens," certain institutional logics (Nigam & Ocasio, 2010), certain mental models (Daniels et. al. 2002;Fortis, Maon, Frooman, & Reiner, 2016;Parmar, 2017;Werhane, 2008). Such models or logics evolve over time, however, and different givens can contradict one another. These processes can give rise to ambiguity and conflicts of power. For example, a company's shareholders and its suppliers might have different opinions on the company's profit goals, and at a certain moment the company's executive will have to think and act in a way that satisfies the expectations of both. An important strategy in potential power conflicts is to talk, question and argue before a decision is taken. Weber and Glynn (2006, pp. 1642-1643 emphasize that sensemaking consists of perceptions, interpretations and actions that can be classified as action formation and aggregation; and this, in turn, introduces institutional elements and power. Managers and stakeholders try to influence each other's perceptions and interpretations, thereby trying to develop collective action, or at least forms of mutual understanding. Weber and Glynn argue that for this reason, sensemaking processes assume institutions-certain mental models-that are related to (1) the roles people play and their expectations related to (2) assumed action/behavior, and (3) frames (that is, typical situations such as a shareholder meeting). These three factors form a mental model with in our with which individuals think and act. For example, many CEOs of big multinationals say that one aspect of their role is to be accountable for corporate policy at the shareholder meeting, and that they therefore listen carefully to the shareholders at the meeting and take action when needed. This can be seen as a key mental model consisting of roles (those of manager and shareholder) in a dialogue (action) in which the management is accountable (frame). In "reality," though, shareholder meetings often appear to be odd gatherings of many people, where many of the decisions under discussion have already been made by shareholders who are not present, but who have voted by proxy. In addition, it is pretty clear how shareholders tend to vote in certain countries. For example, shareholders in the Netherlands voted against just 0.10% of proposals made by the management between 1998and 2002(de Jong, Mertens, & Roosenboom, 2006. This example illustrates the difference between "mental models" of good management and corporate behavior and "reality." Here, the models relate to the roles played by individuals: the way they should interact and the kinds of activities they should organize. Often, for example, a shareholder meeting is defined as a meeting between shareholders and the board in which the latter is called to account for corporate policies. This "model" is different from what an observer would see in practice. A shareholder meeting is not a dialogue between the management and shareholders, but a structured meeting with speeches, some opportunities for questions and a moment at which the people attending the meeting can vote, although the vote is most often decided by the proxy votes submitted beforehand. Given this gap between the mental model of an individual and the dominant mental models in the reality around him or her, enactment can be seen as a bridging component of sensemaking, as part of the process in which individuals (micro) relate to the institutional environment (macro). The observer of the group of people will "decide" that the meeting is a shareholder meeting. This decision gives the meeting a certain degree of relevance and makes sense of the observer's presence there.
Within enactment processes, individuals position themselves in a given situation on the basis of given interests, and are able to reflect on the situation and interests and act on the insights they develop. In this process of constant interaction, or "responsiveness," individuals need to assume a kind of "intermediary position" that enables them to take various interests into account and integrate various perspectives into their actions, whereby their "worldview" is formed. Weber andGlynn (2006, p. 1652) suggest that "enactments set in motion feedback processes that may produce [often-FJdeG] self-fulfilling prophecies that are then used as evidence that environments are objective and external." A CEO's claim that shareholders are only interested in excellent financial performance, for instance, is one example of such an "objective" claim. It not only suggests that shareholders have a onedimensional perspective, but more questionably, it also suggests that shareholders "know" what excellent financial performance is. Behavioral economists have shown that given cognitive limitations, shareholders are unable to evaluate excellent financial performance, but are guided by heuristic approaches such as anchoring (e.g., Ariely, 2008).

| How to make sense within corporate governance
Within a given governance structure, the management has to deal with the interests of various stakeholders. Corporate governance codes and regulations offer little guidance on how managers should do this. Balancing stakeholder interests is a matter of managerial discretion; even corporate governance codes with detailed stipulations on shareholder rights provide little guidance as to the objective of the firm and the role of stakeholders. For example, the Dutch corporate governance code (the Tabaksblat code), which promotes the position of shareholders, describes a firm's objective as follows: The code is based on the principle accepted in the Netherlands that a company is a longterm form of collaboration between the various parties involved. The stakeholders are the groups and individuals who directly or indirectly influence (or are influenced by) the achievement of the aims of the company. In other words employees, shareholders, and other providers of capital, suppliers and customers, but also government and civil society. The management board and the supervisory board have overall responsibility for weighing up the interests, generally with a view to ensuring the continuity of the enterprise. In doing so, the company endeavours to create long-term shareholder value. (Commissie Corporate Governance, 2003, p. 3) This description of a firm's objective is in line with Dutch legislation; the Dutch legislator, in turn, gives little specific guidance as to the objective of a firm. Companies can phrase their own objectives in statutory arrangements. The phrasing they choose, however, can have legal implications for their governance structure. For example, companies that want to implement antitakeover measures should make it clear that their decision making is not dominated by shareholder interests, but should refer to the interests of stakeholders in their statutory arrangements (Van den Berg, 2015). At the board level, similar measures can be taken to ensure that a company is oriented toward its stakeholders. There is a two-tier board system in the Netherlands. The role of the supervisory board is to oversee the executive board. In doing so, the supervisory board should act in the interest of the firm (Burgelijk Wetboek, Boek 2, 140:2, reference to Dutch law), not in the interest of shareholders or another group of stakeholders. Even the supervisory board members who are nominated by the Electronic copy available at: https://ssrn.com/abstract=3587809 works council should serve the interests of the entire company. Various other corporate governance codes use comparable descriptions of the objective of a firm, such as the Cadbury Code in the United Kingdom (e.g., Boatright, 2017;Keay, 2007).
Assuming that Dutch law is not radically different from that of other countries with co-determination, the Dutch example highlights the fact that decision makers enjoy considerable discretion in how they think about and develop corporate principles, given certain institutional conditions. Boards receive little guidance from regulators regarding how to act responsibly, which objectives are most critical, and how various interests should be balanced. They have to formulate their own principles and policies in a complex environment, against the background of a wide range of opinions on what the company should do. Even in a shareholder-dominated country and within a shareholder-dominated theoretical perspective, having a variety of opinions is important for management (e.g., Jensen, 2001).
The fact that the various stakeholder groups and managers have different ambitions leads to a complex mix of ideas and opinions about how a company should develop and what its responsibilities therefore are. The management has to make sense of these opinions in day-to-day policies.
In the complex and ambiguous environment in which a company operates, sensemaking within and outside a company plays a key role in processes of responsiveness in company-stakeholder interaction. The management has to relate conflicting interests to the firm's objective in such a way that the legitimacy of the company is safeguarded, and they do so through processes of sensemaking.
Basu and Palazzo have brought this process perspective back to the heart of CSR research. The sensemaking perspective can be related to value attunement processes within corporate governance (de Graaf & Stoelhorst, 2013). As argued above, enactment is a critical stage in the process, in which actors reflect on their own values in relation to the (assumed) values of others. Corporate representatives have to balance competing views on their social responsibilities. In Swanson's analysis of CSR (1995CSR ( , 1999, value attunement processes are central to a company's interaction with its stakeholders (processes of responsiveness). These processes can lead to changes to a company's principles, policies, and outcomes. Relating these "processes in which CSR mental models develop" to corporate governance perspective allows us to relate individual preferences to organizational and institutional preferences. For example, this could help us to understand the influence of national and institutionalrelated differences in CSR (Blindheim, 2015;Matten & Moon, 2008). Matten and Moon (2008) suggest that the Anglo-American governance system encourages companies to develop more explicit CSR policies, whereas in Europe, more implicit strategies are needed. This perspective also sheds light on the view of the market taken by individuals such as a shareholder-oriented CEO. If individuals adjust their understanding of the market, at a certain moment, this can also have organizational and institutional implications. The impact of these implications is related to the role played by individuals, their position in the governance structure and thereby to their power. This perspective suggests that governance systems play an important role in stakeholder interaction, because they give individuals a degree of power and thereby an ability to shape models of CSR.
Normative notions, the principles of CSR, are institutionalized in governance mechanisms by giving a group of stakeholders influence. Given their position in a governance system, stakeholders can promote certain values and thereby influence regulations and corporate policies. They can also use governance structures to influence corporate principles and policies (de Graaf & Herkströter, 2007). Taking a governance perspective thus gives us more insight into how companies deal with the various expectations of stakeholders, as well as how companies deal with critical tensions between competing views or mental models of the nature of the firm and how a firm and its managers should behave (Matten & Moon, 2008;McWilliams & Siegel, 2001;Mitnick, 2000;Windsor, 2006). Individual decision makers need to interpret and weigh up various interests in order to make a specific decision. These sensemaking processes are inherently ideological in nature (Weick, 1995;pp. 111-113) and require implicit and explicit dialogues (what would they think if I were to do this?), which in turn lead to corporate policies and sometimes to regulation. In recent decades, neoclassical economics and agency theory have played a highly influential role in shaping the mental models of policymakers responsible for governance systems (Djelic & Quack, 2003;Matten & Moon, 2008;Stiglitz, 2004).

PERSPECTIVE ON CSR
In this section, I relate the interaction between individuals, companies, and the institutional environment to the dimensions of sensemaking described by Basu and Palazzo (2008, p. 125). This comes together in Table 1. I then develop three propositions. As we have seen above, sensemaking processes within individuals do not develop completely independent, but are developed in (governance) systems and structures. At the same time, these systems and structures are related to and are part of the mental models of actors. To illustrate this the following example: The governance system consists out of certain laws, but these laws become known by actor who work with the laws, as such they become part of the mental model an actor has of a company. When the law changes, the mental model of the company can change, but next to the cognitive structure of a law, an actor can get other ideas about how a company should look like also, which could influence the development of corporate law on the long run.

| Interaction processes
The governance perspective suggests that individuals represent what companies think; this is a cognitive act. The company itself is "saying" something, for example, in a document that describes the corporate governance structure; this is a linguistic act. The third, conative dimension is related to the behavior of the company in an institutional environment, within its governance system. For example, the company might organize its annual meeting of shareholders, which it is legally required to do, at an odd venue. In the model proposed by Basu and Palazzo (2008), sensemaking processes are relatively fixed, which implies a contradiction. How is it possible for ongoing processes of sensemaking to lead to specific "actions" or activities, such as a shareholder meeting or a works council meeting? Weber and Glynn (2006) propose a way out, suggesting that in every process people play (1) a role with (2) an assumed action related to (3) a certain situation, such as a meeting with stakeholders. Roles, assumed actions, and situations can be described and labeled, although the exact nature of this process changes the moment a researcher starts to describe them.
These value attunement processes (Swanson, 1995(Swanson, , 1999 are shaped by governance systems and structures. As such, the governance system "anchors" the company within its institutional environment, whereby the governance structure relates the stakeholders and managers to the goal of the company, as formulated in the principles of CSR. This goal is intentionally formulated in general terms, giving managers the discretion they need to run the company effectively, and giving stakeholders the ability to influence the company when necessary. Based on this understanding of CSR, we can develop propositions for further research. In conclusion, when we understand CSR as process of sensemaking, we can define three areas of sensemaking in line with Wood's (1991) triple-layered definition of the principles of CSR: at an Role models in which organizations and institutional expectations are related institutional level, the principles of CSR are institutionalized in governance systems; at the organizational level, they are institutionalized in governance structures; and at the individual level, they are institutionalized in the frames or mental models that are made visible through individuals' decisions. CSR in sensemaking processes can thus be related to the three levels of corporate governance, as summarized in Table 1. At each of these three interrelated levels, the principles are open to discussion and are in competition with each other, thereby leading to confusion. There are competing normative ideas at each of these levels and in every area of the economy. Some managers and regulators will be driven by principles relating to agency theory, enforcing transparency and free market mechanisms; others will be more oriented toward their stakeholders, trying to balance various interests and create synergies. Most often, such differences are not openly discussed, but now and then these perspectives compete in the public sphere. Taking a governance perspective can guide the unit of analysis for CSR researchers when focusing on the interaction between principles and regulation, corporate policies, and the motivation of managers. Below, we will further examine this perspective by developing three propositions.

| Making sense of CSR and the role of institutions
Governance systems reflect certain institutional norms and conditions, certain mental models that influence the sensemaking processes of and between the management of the company and its stakeholders. Every actor has a degree of discretion in these processes.
As shown by the quote from the Dutch Corporate Governance code above, principles of "good" corporate governance are phrased in general terms. Likewise, as argued above, companies' principles are also phrased in a general way, leaving room for interpretation. In addition to a general statement about a company's objective (which companies can make more specific in their statutory arrangements and principles), there is also a complex system of regulations to which companies have to respond. This system of regulation is not a stable system and does not offer clear-cut answers for managerial decision making. It has to provide room for managerial discretion, to enable companies to respond properly to developments in their operating context. The context is complex, especially for multinational companies that are active in various jurisdictions. However, companies that are mainly active within a single country also have to operate in a complex and ambiguous environment in which sectoral differences, for example, are critical.
Institutional conditions have a significant influence on these processes. Both the corporate governance literature and the literature on CSR, for example, emphasize the differences between countries with systems of common law and civil law, or network-based or market-based systems. Whereas in market-based countries, such as the United Kingdom and the United States, explicit regulations on corporate principles and policies seem to be dominant, in network-based countries, regulation tends to be principles based (Matten & Moon, 2008;Nooteboom, 1999). In these latter countries, regulation is less specific about what companies should do. Researchers suggest that in network-based countries, companies and their stakeholders have more freedom to interpret regulation to their advantage. This need not influence managerial discretion; a greater degree of regulatory freedom is related to more stakeholder interaction and influence, as is suggested by various researchers (Aguilera, Rupp, Williams & Ganapathi, 2007;de Graaf & Herkströter, 2007;Nooteboom, 1999). In principles-based corporate governance systems, the governance system often offers opportunities for stakeholders to influence the company directly using the influence of the government, the works council or other measures. In countries such as Germany and the Netherlands, various social issues are not addressed via government regulation, but are settled in stakeholder interaction within the governance system.
The government requires the management and boards to take certain actions. Through these institutional requirements, which are conative, the government is able to influence the language of the company and the cognition of managers.
To give an example, the German car manufacturer Volkswagen not only has a works council with far-reaching influence, but the regional government is also a shareholder with decisive power. Given a certain institutional environment (de Graaf & Stoelhorst, 2013;Matten & Moon, 2008), stakeholders will influence organizational principles and policies (Blindheim, 2015;Swanson, 1999) and thereby the mental models of managers. This leads us to Proposition 1.

Proposition 1 When companies are legally obliged to involve a group of stakeholders in decisionmaking processes, this will influence the mental models of the managers of these stakeholders and thereby influence value-attunement processes.
In this proposition, (1) the situation frames (2) an assumed action, influencing (3) the role awareness of participants. This is in line with the ideas proposed by Weber and Glynn (2006), explained in the section above. For example, if managers need to discuss with a works council on a regular basis, they will develop a relationship with the employee representatives. This, in turn, will influence their attitude toward employee representation (e.g., Tapia, Ibsen, & Kochan, 2015;Van der Brempt, Boone, Witteloostuijn, & Berg, 2015). In order to survive in a system in which employee representation is seen as important, managers' behavior has to be consistent with the normative notions underlying the governance system. This could even lead to changing attitudes and mental models, for example, when a manager experiences his relationship with the works council as constraining or advantageous. A survey of managers' attitudes toward forms of stakeholder influence could help to further our understanding of the relation between attitudes toward stakeholders and the ways in which stakeholder interaction is organized. The relation that we are suggesting here is that conative patterns (a certain action, such as a weekly meeting) lead to certain linguistic patterns that influence the cognition of individuals. Assuming interaction patterns, individual managers and stakeholder representatives could demand certain forms of corporate behavior that could become "conative;" as such, they could become part of intended behavior, the institutional practices of a given time. For example, in the Netherlands it is impossible for companies to oppose unions and works councils, given their institutional role, whereas in the United States, regulation offers more opportunities for antiunion policies, and works councils as such do not exist.

| Making sense of CSR and the position of organizations
The organizational perspective has long been dominant in the CSR literature. From this perspective, CSR is part of an organization's strategy (e.g., Wood, 1991Wood, , 2010. Managers have to deal with stakeholders in ways that are beneficial to the organization. In order to do so, they assess the environment, engage with stakeholders and manage issues, all in relation to and in interaction with corporate principles (Wood, 1991). In practice, the linguistic expression of these principles is captured in mission statements, codes of conduct, and business principles.
The institutional perspective on CSR as a process underlines the importance of sensemaking processes within a company and within the institutional environment (Blindheim, 2015). In this view, given a certain governance structure, companies can deviate from their environment (de Graaf & Stoelhorst, 2013;Filatotchev & Nakajima, 2014). The company's representatives need to have a similar mental model of the company's role in society, and this model can and often must deviate from general societal expectations of companies. For example, a company that produces for a niche market can deviate further from public expectations than a supermarket that is active in every part of the country.
To enable corporate representatives to deviate from general expectations, certain features are needed; certain acts that demonstrate that "we do it differently." These can take the form of a mission statement or an approach to the market, but they also need to be reflected in the governance structure (Bezemer et al., 2015;de Graaf & Stoelhorst, 2013), otherwise complications can emerge. This leads us to Proposition 2.
Proposition 2 In a given governance system, specific measures in the governance structure enable managers to deviate from dominant mental models in their decision-making on CSR.
In this case, the organization acts as a moderator, giving a degree of direction to expected actions and behavior. The organization does not only "bring messages" from the environment to managers, of course; executives can also use the organization to promote certain practices at the institutional level. This proposition is consistent with Barnett (2014), who argues that organizations can "say" something while doing something else, and can even do something while managers think they are doing something else. The proposition explains the dynamic between (1) assumed actions within an organization, given the role of managers (2) in (3) a certain situation, and is thereby also in line with Weber and Glynn (2006).
Regulation plays a role in facilitating behavior by companies that differs from the dominant institutional logic. Two examples from the Netherlands demonstrate this point. First, antitakeover measures are not common practice under Dutch legislation. A company is only allowed to implement certain antitakeover measures when referring to a stakeholder orientation in its statutory arrangements. The second example concerns the works council, a common institution in the Netherlands. However, Dutch companies are not obliged to have a works council at the corporate level when the company is mainly active outside the Netherlands. Wood (1991) relates managerial discretion to the principles of CSR. An individual has to deal with his own ideas about how an organization should be managed. In his decision making, he is tied to principles at the institutional and organizational level. Managers act in an environment in which various, often conflicting, opinions attempt to dominate.

| Making sense of CSR and the position of individuals
The governance system in which a manager operates and the organization's governance structure enables the manager to make sense of conflicting opinions and related interests, certain mental models thereby become dominant. In a shareholder-dominated environment, the manager's mental model will not only be influenced by shareholder interests (and his assumptions about these interests), but also employees and clients will refer to shareholder interests when promoting a measure that they consider to be in their interest.
Once we understand CSR as a process of sensemaking, we gain new insights into the relationship between the individual manager and the institutional environment. Palazzo and Scherer (2008) emphasize the role played by mental models in CSR. They argue that research has long focused on CSR activities, where the institutional environment and the mental models that are developed in this environment play critical roles in decision making. If we relate this perspective on sensemaking to governance regulation, the interaction between decision makers and how they understand (cognitive) their role, their organizations and institutional expectations about behavior become apparent (e.g., Berthoin Antal & Sobczak, 2014). In this case, individual cognitive understanding has to be in line with the language of companies and expected behavior (e.g., Barnett, 2014). This leads us to Proposition 3.

Proposition 3 When a manager's individual sensemaking of CSR is out of step with dominant men-
tal models, the position of the individual within the governance structure of a company will be adjusted.
Research on key cases could be used to develop this proposition further, which is in line with Maitlis and Sonenshein (2010). It is interesting to note that key incidents, such as the takeover can have a huge impact on subsequent decision making by managers, politicians, and other policymakers. After the takeover the attitude of stakeholders toward a company can changesignificantly, which can lead to clients or employees that run away and new government regulation.
Taking a governance perspective on CSR thus allows us to relate individual decision making to institutional characteristics. The manager is not only a decision maker, but he also represents certain values (e.g., Maitlis, 2005) and principles, a role that can entail much more than simply representing decisions that have been made within a particular managerial remit. We could develop this proposition further by assessing how managers function as role models and their impact on business and society (e.g., Grant, Dutton, & Rosso, 2008).

| DISCUSSION
The governance perspective on processes in which mental models of CSR develop, suggests that CSR researchers have become hostage to a narrow, economics-based perspective in order to try to prove the financial added value of CSR (Margolis & Walsh, 2003). In the recent discussion about the economic relevance of CSR, CSR researchers appear to have become more self-aware. The "S" of CSR is crucial for our understanding of the dynamics of companies and economies (Aguilera et al., 2007). Ethics belongs at the heart of the marketplace, argue Purnell and Freeman (2012), and no separation should be made between economic rationality and social and ethical notions (Harris & Freeman, 2008). Every CSR perspective that focuses on one or a few groups of stakeholders is thereby making a political and an ideological statement.
The political, normative side of economics found its biggest proponent in Milton Friedman, known for his mental model that "the business of business is doing business" (1962, p. 9). Corporate governance regulation has been a critical arena for this political agenda (Veldman, 2013), in which agency theorists have played a prominent role (de Graaf & Williams, 2009). Inspired by Friedman and other economists, agency theorists argued that companies and their managers should have one objective, that of acting in the financial interests of the shareholders (e.g., Jensen & Meckling, 1976). It was claimed that by achieving this single and clearly measurable financial objective, other stakeholders would also be served (Jensen, 2001). The political and social complexity in the relationship between business and society disappeared from their models. To counter the criticisms of economists and make a business case for CSR, researchers have tried to develop complex metrics to measure the financial importance of CSR (Margolis & Walsh, 2003;Orlitzky, Schmidt, & Rynes, 2003). They have also tried to prove that companies need CSR in order to achieve their instrumental objectives.
If we view CSR in corporate governance from a sensemaking perspective, we gain a better understanding of why CSR is difficult to define and measure. Both countries and companies have certain leading principles that are normative by nature, develop over time, and differ between companies and countries. The context-dependent character of CSR, which assumes the involvement of all stakeholders, makes it a multifaceted concept. The governance perspective shows us that within dialogue mechanisms, principles become institutionalized in measures that are context dependent (Bovens, 1998), related to a specific institutional context (governance system) and a specific company (governance structure). Not only is the quality of the outcomes critical for stakeholder influence, as is argued by agency theorists, but also the quality of the processes that lead to the outcomes, as is argued in the CSR literature (e.g., Bauer & Ackerman, 1976;Wood, 1991).
This puts power relations, and thereby corporate governance, at the heart of CSR research. If there is no dominant economic rationale by which to identify the most appropriate course of action, other mechanisms have to play a role in decision making. This makes the power of an individual or a group more relevant. CSR's political nature also explains why it has to be underdefined, like concepts such as "welfare," the "continuity of the firm," "fiduciary duty," or "act in the interest of the company." All of these notions are critical in our mental models of modern business systems and enable economic actors to make the best possible decisions in a given situation. In this view, CSR is and should be difficult to measure. When CSR is no longer context dependent, it is no longer possible for economic actors to take responsibility, managerial discretion would not exist.
Acceptance of the context-dependent character of CSR suggests a number of new avenues for research. Dialogue in governance systems and governance structures can be studied to identify mental models and patterns in their development (e.g., Haigh & Griffiths, 2012). In interaction processes, the conative, linguistic, and cognitive dimensions of CSR merge. By taking a descriptive microeconomic approach, as suggested by Margolis and Walsh (2003), researchers might further their understanding of how certain managerial values or motivations are shaped in certain structures, as well as of how interest groups try to change certain structures and "revalue them," new models come into existence. The real or perceived contradictions between what companies say, what they do, and how this can be understood, can also be assessed. Research has already been done into how various actors behave in certain governance systems (e.g., Matten & Moon, 2008;Whitley, 1999). Some authors have even considered the implications for the social role of companies. CSR research could benefit from such insights, especially when attention is paid to the role of power in stakeholder interaction. From a governance perspective, principles of CSR lead to dialogue in which various stakeholders exercise different forms of power, as well as to regulative measures in governance systems and governance structures. These systems and structures, in turn, influence the dialogue on the principles.
From a governance perspective, mental models guide organizations and individuals and function, as suggested earlier, as the bed of a river in which there is a constant flow of opinions (e.g., Weick, 2012). This perspective requires more interpretive studies on CSR that draw on insights from law, in which interpretive methodology plays a standard role in practice and academic theory. From both a methodological standpoint and with a view to empirical research, this would lead to challenging new research avenues in the area of law and economics. One interesting question is whether this would call for a social constructivist perspective on CSR alone (Fortis et al., 2016).
By assessing CSR as a sensemaking processe in corporate governance structures and systems, we see that CSR is not the result of managerial activity alone. Basu and Palazzo (2008, p. 124) define CSR as a managerial process in which relationships with stakeholders are considered and discussed, and the position and work of managers is related to the common good. Taking a corporate governance perspective makes clear that stakeholders are also partly responsible for corporate behavior and the common good.
Adopting a governance perspective facilitates an understanding of how power in relationships can shape, through a process of dialogue, principles of CSR in regulations and corporate policies. For example, even when a CEO is no longer the leader of a company, his opinion is related to his former | 445 de GRAAF position there. This makes his opinion more "important" and more powerful than that of another ex-employee, although this is not to suggest that the values of a former CEO are always attuned to and in line with the dominant opinions in his institutional environment. Having a powerful position can also allow one to influence a country's governance system. Given their past positions, former CEOs or other corporate executives are sometimes appointed to committees that revise regulations on corporate governance, thereby allowing them to influence the governance system of a country. Here, CSR research could benefit from research on corporate governance, such as the research from a behavioral perspective (e.g., Van Ees, Gabrielsson, & Huse, 2009). This might enable researchers to further their understanding of the limits of judgment in corporate decision making, and the impact that this has on our understanding of a complex theme such as CSR.
The governance perspective leads us to consider questions that make comparative research on corporate systems and multinational companies very relevant. As Scherer and Palazzo (2010) argue, companies operate in a complex political environment in which no single jurisdiction plays a decisive role. Many companies work in various jurisdictions that have different expectations about good corporate behavior. Various European companies, for example, have faced serious penalties in the United States for corruption, despite facing no legal problems in their home country and being backed by their "own" governments.
Although the sensemaking perspective on CSR does appear relevant, it has not yet provided us with an answer to questions about managerial accountability. The argument proposed by Jensen (2001) and others, that a stakeholder perspective makes managers less accountable for their policies, still holds. If complexity, ambiguity, and stakeholder interests are so critical in managerial action, is managerial accountability still possible? More attention could be paid to "voice," stakeholder interaction, in research and practice. Stakeholders are often well aware where the "good role models" are to be found. This can also help to develop "good" role models. Reputation seems to be an important control mechanism in networks (Jones, Hesterly, & Borgatti, 1997). Research on Proposition 3, in particular, could shed more light on the role played by individuals and mental models.

| CONCLUSION
In this article, I have argued that CSR as sensemaking process is determined by how certain normative notions are institutionalized by governance systems and governance structures. The interaction between principles of CSR and an institutional framework or governance system-something that is mainly reflected in regulation-therefore plays a critical role in CSR research. Influence pathways thus deserve more attention, on the grounds that they establish sensemaking processes.
Owing to the fact that CSR is under-defined, decision makers need to form normative judgments about developments in their environment and relate these judgments to a company's principles. These judgments are "tested," value attunement takes place. As part of these sensemaking processes, CSR becomes institutionalized in dialogues that are structured in systems and structures. In these interactive processes, normative notions such as shareholder dominance can become part of a company's instrumental objectives regardless of whether this is a smart strategy for the company in the long term. If these interaction processes were not structured, managers would lack grounds on which they develop their mental models and make decisions. By contrast, if the expected outcomes were clearly defined, value attunement would not be possible in sensemaking processes and corporate behavior would be predetermined.
This implies that CSR will always be difficult to measure, but it can be understood by assessing power structures and sensemaking processes within these structures. From the position of stakeholders in a governance system and a governance structure, the normative notions that guide corporate behavior are institutionalized. When stakeholders want to assess CSR, they should take into account the power of stakeholders within interaction processes, within systems and within structures. For this reason, CSR research needs to be clear about the underlying assumptions about corporate governance, and corporate governance research needs to be clear about the dominant mental models on CSR that influence the outcomes.