Overcoming the Biases of Microfoundationalism

1. Mechanism-Based Explanations, Social Mechanisms, and Microfoundationalism in Analytical Sociology

Although analytical sociology is not a unified doctrinal system, nor is there any single founding father or mother of this movement, there nevertheless are certain methodological ideas that unite analytical sociologists. In his account of the key principles of analytical sociology, Hedström (2005, 1) writes that analytical sociology “seeks to explain complex social processes by carefully dissecting them and then bringing into focus their most important constituent components.” Analytical sociologists thus aim to deliver explanatory understanding about social macro-phenomena by developing precise, abstract, and analytically realist explanatory theories that refer to social mechanisms that causally generate these phenomena (Hedström 2005, chap. 1).

Hedström (2005, 2) further specifies the mechanism approach to the explanation of social phenomena as follows:

The core idea behind the mechanism approach is that we explain a social phenomenon by referring to a constellation of entities and activities, typically actors and their actions, that are linked to one another in such a way that they regularly bring about the type of phenomenon we seek to explain.

Insofar as social mechanisms that bring about social phenomena are assumed to be “recurrent processes linking specified initial conditions and a specific outcome” (Mayntz 2004, 241) (rather than abstract theoretical models or parts of theories), it can be said that mechanism-based explanations explain by opening up the black boxes that connect the causes to their outcomes (Elster 1989, 3-10). In other words, “proper explanations should detail the cogs and wheels of the causal process through which the outcome to be explained was brought about” (Hedström and Ylikoski 2010, 50).

Mechanism-based explanations of this kind are different from covering-law explanations, since the former do not usually involve any general law statements nor is the deductive form necessary to mechanism-based explanations (e.g., Hedström 2005, chap. 2; Hedström and Ylikoski 2010, 54-55). Due to the fact that social mechanisms are typically assumed to consist of social actors and their activities, mechanism-based explanations are also regarded as alternatives to functionalist and structuralist explanations that abstract from concrete social actors and their activities. Moreover, the above views are meant to differentiate the explanatory methodology of analytical sociology both from “grand” social theories, which are deemed too general, abstract, and ambiguous for explanatory purposes, as well as from the tradition of correlation-based statistical “causal modeling,” which is criticized for its eclectic empiricism and tendency to replace real social actors with statistical variables (e.g., Hedström 2005, chaps. 1 and 5). But what exactly are these social mechanisms that are said to bring about social phenomena?

A good place to start answering this question is to look at Hedström and Ylikoski’s (2010, 50) characterization of “the general ideas that are shared by most accepted definitions [of the concept of mechanism—T. K.]” in the current philosophy of science:

1. [A] mechanism is identified by the kind of effect or phenomenon it produces. A mechanism is always a mechanism for something [. . .].

2. [A] mechanism is an irreducibly causal notion. It refers to the entities of a causal process that produces the effect of interest. [i.e., mechanism approach to causal explanation presupposes causal realism—T. K.]

3. [T]he mechanism has a structure. When a mechanism-based explanation opens the black box, it discloses this structure. It turns the black box into a transparent box and makes visible how the participating entities and their properties, activities, and relations produce the effect of interest.

4. [M]echanisms form a hierarchy. While a mechanism at one level presupposes or takes for granted the existence of certain entities with characteristic properties and activities, it is expected that there are lower-level mechanisms that explain them. [. . .] The only requirement [for a mechanism to be explanatory—T. K.] is that such entities, properties, and activities really exist; their explanation is a separate question [which means that there is no regress of explanations involved here—T. K.]. (Hedström and Ylikoski 2010, 50-52)

Now, these assumptions give us some idea what causal mechanisms are taken to be in analytical sociology since recent accounts of social mechanisms in this movement are influenced by the new mechanistic philosophy of science.

For example, drawing explicitly on Machamer, Darden, and Craver’s (2000) influential paper on mechanism-based explanations, Hedström (2005, 25) writes that

mechanisms can be said to consist of entities (with their properties) and the activities that these entities engage in, either by themselves or in concert with other entities. These activities bring about change, and the type of change brought about depends upon the properties of the entities and the way in which they are linked to one another. A social mechanism, as here defined, describes [is?—T. K.] a constellation of entities and activities that are organized such that they regularly bring about a particular type of outcome. We explain an observed phenomenon by referring to the social mechanism by which such phenomena are regularly brought about.

I think that Hedström’s account of mechanisms fits quite nicely to Mayntz’s (2004, 241; also Hedström and Ylikoski 2010, 50) characterization according to which “mechanism statements [or models and theories that refer to mechanisms—T. K.] are causal generalizations about recurrent processes.” As was already indicated above, many analytical sociologists assume a position of causal realism, which includes an assumption that “causal connections between events are real and are conveyed by the powers and properties of entities” (Little 2011, 273), although analytical sociologists rarely specify their account of causal realism or their social ontological assumptions.

What is important to the topic of this article is that Hedström (2005, 28-29) and many other analytical sociologists (e.g., Demeulenaere 2011; Elster 1989) make a further assumption according to which social mechanisms consist of interrelated human individuals and their activities. Hedström and Bearman (2009, 4) make this point clearly when they write that mechanisms explaining social facts “invariably refer to individuals’ actions and the relations that link actors to one another.” Hedström (2005, 115; also 19, 70-74) accordingly contends that, when we are modeling social mechanisms that bring about social macro-phenomena, we must assume that “it is individuals, not social entities, that are endowed with causal powers.” Individualist social ontology is presupposed, too, in Hedström and Swedberg’s (1996, 299) bold statement that “there exist no macro-level mechanisms.” Given his commitment to ontological individualism, it is then no wonder that Hedström (2005, 73) ends up contending that “I do not believe that ontological collectivism [i.e., non-individualist social ontology—T. K.], in whatever form or shape, has anything to offer sociological theory.”

The above quotations from the programmatic papers by analytical sociologists clearly indicate that analytical sociology is firmly rooted in the traditions of methodological individualism and microfoundationalist thinking (see also Demeulenaere 2011; Little 2012). Some analytical sociologists have nevertheless rejected the strong (or traditional) version of methodological individualism, and relied instead on a weak version of this doctrine termed as structural individualism. Hedström and Bearman (2009, 8), for example, define the latter position as “a methodological doctrine according to which all social facts, their structure and change, are in principle explicable in terms of individuals, their properties, actions, and relations to one another.” This position is said to differ from the traditional methodological individualism in that it emphasizes “the explanatory importance of relations and relational structures” (Hedström and Bearman 2009, 8; cf. Coleman 1990). In addition, current analytical sociologists explicitly reject all atomistic conceptions of human individuals and consider human individuals as social beings, whose properties and actions are influenced by their interactions with other people (e.g., Demeulenaere 2011; Manzo 2010).

On the basis of the views reviewed above, I suggest that the position of IMF can be defined as comprising the following four theses:

All macro-level social phenomena are ultimately brought about by the interconnected actions of interrelated individuals.

The first two of these theses are ontological, the third is methodological prescription, and the fourth is the combination of these two types of thesis. I think that my formulations of the three first theses are relatively uncontroversial (at least in the context of analytical sociology), but perhaps the fourth one should be discussed a bit further. One reason for this is that Barbera (2012, 3) has recently invoked the distinction between causally generative properties (or causal powers) and causally relevant properties in this context. He argues that, although macro-level properties are not causal powers of social entities, they may nevertheless be causally relevant in explanations of macro-level outcomes of interactions of many individuals due to the fact that they “influence, affect, shape, impact, and concern” (Barbera 2012, 3) the actions of relevant individuals. If this is true, then (in contrast to thesis 4) macro-level properties would not be causally inert even if they are not emergent causal powers of social entities.

I nevertheless doubt whether Barbera’s (2012) distinction between causally generative and causally relevant properties is sound. First of all, it can be argued that the “causal relevance” of some macro-level properties (e.g., distributions of properties of individuals in a population) can be explicated in terms of the aggregates of the causal powers of individuals that compose the population. Insofar as this is the case, the previous distinction is redundant. Furthermore, it can be argued (in contrast to IMF) that many other macro-level properties, such as social norms, institutions, and structures of social networks, actually presuppose the existence of social entities of whose emergent causal powers and properties these norms, institutions, and relational structures are (see Elder-Vass 2010; Kaidesoja, 2013). In this case, too, the distinction between causally generative and causally relevant properties of social entities evaporates, since causal relations that involve social entities are conveyed by their generative causal powers (as well as causal powers by other relevant entities). ³

What is important to the ensuing arguments is that proponents of IMF assume by definition that all causal powers ascribed to social groups are ontologically reducible to the aggregates of causal powers of their individual members, which implies that social groups cannot be considered as collective agents. Next, I ask whether the concept of social mechanisms and the model of mechanism-based explanation really presuppose IMF.

2. Does the Mechanism Approach to Social Explanation Presuppose IMF?

If we consider the above list of the general ideas associated to causal mechanisms and Hedström’s account of the concept of social mechanism, then we can see that all that is required by them is that social macro-phenomena under explanation are brought about by interacting entities (or actors) of some kind that exist at the lower level of organization with respect to the macro-phenomenon to be explained. As also several other authors have indicated (Kincaid 1996; Mayntz 2004; Vromen 2010; Wan 2011, 2012; Ylikoski 2012), nothing in the mechanism approach to explanation as such necessitates that causally efficacious entities referred in the mechanism-based explanations of social macro-phenomena should always be individual human beings. Actually, the idea that mechanisms form a hierarchy (see item 4 in Hedström and Ylikoski’s 2010 list) can be used to question the very requirement according to which explanatory social mechanisms should always consist of interacting individual human beings: If it can be shown that there are relatively enduring collective agents with characteristic emergent capacities and activities, then there is no reason why the social entities of this kind could not form parts of social mechanisms beyond the level of interacting individuals. These points then imply that IMF is not necessarily presupposed by the mechanism approach to social explanation.

It is interesting to note that in the context of biological research, where mechanism-based explanations are common, mechanism models often incorporate references to biological entities with system-level capacities (i.e., weakly emergent powers) at many different levels of organization. Examples of such entities include organic molecules, organelles, cells, organs, organisms, and structured groups of organisms (see Machamer, Darden, and Craver 2000; Wimsatt 2007). Hence, it is not required in these explanatory practices that all mechanism-based explanations should be developed in terms of some fundamental level of causation, nor is it assumed that there is some fundamental level of causally efficacious biological entities. Rather, entities with capacities to engage in productive activities are assumed to exist at different levels of organization in biological systems. This is because

Higher-level entities and activities are [. . .] essential to the intelligibility of those at lower levels, just as much as those at lower levels are essential for understanding those at higher levels. It is the integration of different levels into productive relations that renders the phenomena intelligible and thereby explains it. (Machamer, Darden, and Craver 2000, 23)

Explanations that refer to entities (with capacities) existing at different levels of organization nevertheless presuppose that the system-level capacities (or weakly emergent causal powers) of those entities can be explained in terms of the properties of their lower-level components, the relations, and interactions of these components, as well as (at least in some cases) in terms that describe the entity’s interactions with its environment. Nevertheless, empirically established inter-level reductive explanations of this kind eliminate neither of these entities nor their causal capacities from the biological ontology, but rather provide empirically grounded reasons to count them among the core entities in mechanism-based explanations of higher-level biological phenomena (Wimsatt 2007, chap. 12).

These views, then, question the link between IMF and mechanism-based social explanations, since they cast serious doubt on the claim that there has to be a fundamental level of causally efficacious entities in social reality. However, to assess the merits of the mechanism approach to social explanation that assumes IMF, we must evaluate its outcomes in the context of empirical research as well as consider the plausibility of the ontological assumptions underlying this position. I turn next to these questions.

3. Biases of IMF in Explanatory Research

In her perceptive analysis of the recent literature on social mechanisms and mechanism models in the social sciences, Mayntz (2004, 255) contends that

The problem is that our theoretical tool boxes for different types of [social—T. K.] mechanisms are very unevenly filled. We already have a good-sized, if not very orderly, tool box of mechanism models for different forms of collective behavior—collective in the sense that the uncoordinated, but interdependent, actions of many individuals generate aggregate effects. Examples are models of linear and nonlinear diffusion, the mechanism underlying spatial segregation in urban housing, the market mechanism, and the mechanism of mobilization where not only thresholds but also a “production function” (i.e. how many must participate to produce the effect) plays a role. [. . .] We have as yet no similarly filled tool box for mechanisms where specific types of corporate actor constellations and relational structures play the crucial role. (italics added)

I agree with Mayntz’s observation that most of the theoretical (or formal) models about social mechanism that have been proposed in sociological literature aim to explain social macro-phenomena by explicating how “the uncoordinated, but interdependent, actions of many individuals generate aggregate effects” (for similar views, see Baldassarri 2009; Little 2012). This also means that all the examples of mechanism models mentioned by her are compatible with the doctrine of IMF. In addition to these examples, theoretical models on dissonance-driven desire formation (Festinger), rational imitation (Hedström), and self-fulfilling prophecies (Merton), all of which have been discussed and formalized in analytical sociology, are compatible with both Mayntz’s point and the position of IMF.

Given that analytical sociologists typically combine the mechanism approach to social explanation with IMF, the absence of theoretical tools for analyzing macro-mechanisms is not at all surprising. This state of affairs could be seen rather as an implication of the application of their methodological prescriptions. In other words, the combination of the mechanism-based model of explanation and IMF suggests that “there exist no macro-level mechanisms” (Hedström and Swedberg 1996, 299), and for this reason, “all social facts, their structure and change, are in principle explicable in terms of individuals, their properties, actions, and relations to one another” (Hedström and Bearman 2009, 8). Now, due to the fact that they deny the existence of macro-mechanisms, the advocates of IMF might reply to Mayntz by claiming that there are no methodological biases involved here since the content of our theoretical tool-box reflects the true nature of social reality. I will challenge this latter claim below by arguing, first, that macro-mechanisms are referred to in some interesting social scientific theories and explanations. Second, I will try to indicate that, once we take into account some recent approaches in cognitive science, especially the distributed approach to human cognition, the notion of collective agent with emergent causal powers is not at all as problematic as the proponents of IMF tend to assume.

To be fair, there are some analytical sociologists (e.g., Baldassarri 2009; Barbera 2012; Barkey 2009; Demeulenaere 2011; Manzo 2010) as well as some predecessors of this movement (e.g., Coleman 1990) who have recognized the need for such theoretical models that take explicitly into account not only the distributions of properties (in populations of individuals) and the structures of social networks but also social norms, institutions, and relatively enduring organizations. With some exceptions (e.g., Baldassarri 2009), the concepts of social relation and network structure used by many analytical sociologists, however, appear to be rather thin in that their theoretical models tend to downplay the significance of the socially coordinated, organized, and institutionalized forms of social relations and interactions. Moreover, when social norms, institutions, and culture are taken into account by analytical sociologists, there is a tendency to ontologically reduce them to the aggregate properties of groups of individuals (e.g., shared or typical beliefs and expectations of individuals). For example, Hedström’s (2005, 69) analysis of social change and interaction explicitly focuses on “the social patterns of desires, beliefs and actions that are likely to emerge when large number of individuals” interact affecting the beliefs and desires of each other. Similar position is presupposed by Elster (2009, 196) when he writes that “[a] social norm is simply a shared expectation [in a group of individuals—T. K.] that others will react to a given behavior in a way that is painful to oneself.” Nothing prevents one claiming that the patterns of shared beliefs and mutual expectations of the members of a group comprise the culture, norms, and institutions of that group, but the problem is that this is a very narrow and individualist understanding of these properties, which I think can be (and has been) questioned with good reasons (e.g., DiMaggio 1997; Little 2012; Santoro 2012). So I would say that the ideas related to IMF are still very much alive among the prominent analytical sociologists though I admit that they are not all shared by all researchers who associate themselves to this methodological movement (see Baldassarri 2009; Barbera 2012; Barkey 2009; Manzo 2010; also Wan 2012, 1552-53).

It is interesting to note that the methodology based on IMF is clearly incompatible with many of the methodological ideas and explanations developed in recent historical and comparative macro-sociology, including research on political processes. This is because socially coordinated and institutionalized actions of individual actors as well as actions of organized social groups (e.g. social movements and formal organizations) and larger social structures are typically among the core explanatory factors in the explanations of large-scale macro-social changes, such as state formation, rise of nationalism, democratization, revolutions, wars, globalization, or emergence of social policies (e.g., see Barkey 2009; Held and McGrew 2007; Little 2012; Mahoney and Rueschmeyer 2003; McAdam, Tarrow, and Tilly 2001; Smith 1991; Stinchcombe 1998; Tilly 2001, 2008). A research program on contentious politics also explicitly recognizes the importance of social mechanisms beyond the individual-level for explanations of political processes (McAdam, Tarrow, and Tilly 2001; also Tilly 2001, 2008). Moreover, there are some theoretical models about social mechanisms developed in historical sociology and in organization studies in which (at least some of) the core actors are formal organizations. Examples include Stinchcombe’s (1998) model of the monopolist competition between corporate groups (e.g., business corporations, universities, and nation-states) in competitive fields (e.g., markets, prestige systems, and the world power system) that involve field-specific rankings in terms of profits, prestige, or power. Hannan and Freeman’s (1989) and their followers’ theoretical models about various selection mechanisms in organizational populations are also cases in point.

Now, unless one is able to show that social explanations and mechanism models that break the assumptions of IMF are always worse in terms of explanatory power than their alternatives based on IMF, then it should be concluded that the latter type of methodology is problematically biased. Next, I will take a closer look at some plausible candidates for the social macro-mechanisms that have collective agents as their components.

4. Candidates for Macro-Mechanisms

The research program of contentious politics is a rich source of ideas on meso- and macro-level social mechanisms (e.g., McAdam, Tarrow, and Tilly 2001; Tilly 2001, 2008). Some of the following candidates for macro-mechanisms, such as coalition formation and competition, are mentioned in the contributions of the proponents of this research program. I will not, however, explicitly draw on these contributions. There are two reasons for this: First, the account of social mechanism in terms of events and the distinction between mechanisms and processes, presented by McAdam, Tarrow, and Tilly (2001, 24), are not entirely compatible with the above view on social mechanisms. Second, many of the ideas on different types of social mechanisms put forward in this program are rather sketchy in that they do not always specify the nature of entities and activities that are thought to bring about change in each social mechanism (cf. Little 2011). Therefore, I will focus here exclusively on the macro-mechanisms that are compatible with Hedström and Ylikoski’s (2010) account of causal mechanisms.

Here is my list of candidates for social macro-mechanisms that include collective agents among their core components:

competition between collective agents (e.g., business firms competing for market shares in competitive markets, universities competing for bright and wealthy students in a higher education system, nation-states competing for political power in an international power system),

It is important to bear in mind that the items in this list refer to abstract types of macro-mechanisms (or mechanism schemes), each of which is thought to bottom out to subspecies consisting of concrete macro-mechanisms. I also believe that this list could be easily extended and that each subspecies of the list could be further specified, or even modeled theoretically, by relying on the relevant research literature.

In every concrete macro-mechanism the core actors are collective agents, or both collective agents and human individuals—not just individuals. This is because it is not possible, or so I will later argue, to ontologically reduce the cognitive capacities of collective agents involved in these mechanisms to those of their individual members or the aggregates of the latter. Drawing on the emergentist systemist ontology of causal powers developed elsewhere (Kaidesoja, 2013; Wan 2011; cf. Bunge 2003), collective agents, such as formal organizations, can be understood as concrete social systems ⁴ with emergent causal powers, relational structures, and mechanisms underlying their agency. To my mind, this view on social macro-mechanisms is intuitively plausible, since, for example, competition between business firms in competitive markets does not seem to be ontologically reducible to the interactions between members of different firms for the simple reason that it is precisely the emergent capacities and competitive activities of the firms in markets that are essential constituents of this mechanism (cf. Stinchcombe 1998). This is not to deny, however, that competitive actions of each environmentally embedded firm are ontologically dependent on the materially mediated interactions of their personnel in an analogical way as the intentional actions of human individuals are ontologically dependent on the interactions of their environmentally embedded brains and bodily organs.

It should be stressed that each of the collective agents in macro-mechanisms sketched above is embedded in a certain institutional environment. Indeed, institutional environments play a crucial role in many historical explanations that refer to macro-mechanisms and social processes that take long periods of time to unfold, since these environments can have a significant effect on how the mechanism or process of interest unfolds in various contexts (e.g., see Pierson 2003). In my view, institutional environments of this kind can often be explicated in terms of some larger social system (e.g., capitalist market economy, nation-state, or international power system) of which the collective agent forms a part.

It is also worth adding that the ascription of agency to a structured social group does not amount to a denial that, at least in more or less democratically organized groups, its individual members still have many degrees of freedom in their actions both as members of these groups as well as in other spheres of their lives unrelated to their membership in this particular group. Indeed, as I will later suggest, collective agency can be usefully considered as a concept that allows degrees since some organized social groups (e.g., formal organizations) are clearly more unified, cohesive, and enduring entities than others (e.g., fleeting interaction groups).

Although macro-mechanisms are occasionally referred to in explanatory macro-sociology, their significance should not be overrated, since not all social mechanisms relevant to the explanations of social macro-phenomena involve interacting collective agents. It thus depends on the explanatory question and the macro-phenomenon to be explained whether references to macro-mechanisms are needed. Moreover, postulations of macro-mechanisms in mechanism-based explanations of social phenomena should be done with care. One reason for this is that sometimes sociologists misattribute agency to theoretical abstractions, or taxonomic concepts, such as “capital,” “nationalist ideology,” “bourgeoisie,” “working class,” or “discourse of governmentality,” which do not refer to such concrete social systems that may be said to possess emergent causal powers ⁵ (see Kaidesoja 2007, 2013). Hence, in explanations of social macro-phenomena, claims about the actions and interactions of collective agents should always be substantiated by empirical evidence that shows that such organized social systems that can be plausibly understood as collective agents (with emergent powers) were present in the causal process of interest.

It should be emphasized again, however, the requirement that claims about actions and interactions of collective agents should be backed up with empirical evidence does not mean that every explanation of large-scale social macro-phenomena has to explicate the individual-level microfoundations of these actions and interactions, and/or explain the historical origins of the collective agents of interest, in terms of interrelated and interacting human individuals. By contrast, in many explanatory studies on large-scale macro-phenomena, it is sufficient that we have a general understanding how the collective agents of this kind function (e.g., how collective-decisions are typically made in the organizations that are the components of the relevant macro-mechanism) and empirically grounded reasons to believe that the macro-phenomenon of interest was causally generated by the interactions of this kind of collective agents with emergent powers (cf. Little 2012). The explanation of the emergent powers of the relevant collective agents is simply a separate explanatory question. Of course, it is always possible to zoom in to a particular collective agent and study the underlying mechanisms of its emergent causal powers, but this type of research requires the uses of different methods and data from the explanatory studies on large-scale macro-phenomena. Note that the situation is largely analogical with respect to social mechanisms that consist of interacting individual agents with emergent causal powers (e.g., capacity to act intentionally and understand intentions of others), since explanations of social phenomena that refer to mechanisms of this kind typically take for granted the existence of certain causal powers of individuals while assuming that these powers can be explained in reference to their underlying cognitive and neural mechanisms (cf. Hedström and Ylikoski 2010). In the next section, I suggest that the distributed cognition perspective combined with organization studies enables us to get a better grasp of the underlying mechanisms of the emergent cognitive capacities of collective agents.

5. Collective Agents as Distributed Cognitive Systems

Let me begin with a brief characterization of the concept of cognition presupposed in much recent cognitive science, including the perspective of distributed cognition. Following Theiner and O’Connor (2010), this can be termed as a big tent approach to cognition. According to this approach, the concept of cognition is a cluster concept that “subsumes a more or less loosely knit family of capacities that we can distinguish for taxonomic purposes” (Theiner and O’Connor 2010, 82). A system is then taken to be cognitive to the extent that,

It can adapt its behavior to changing environments.

Concrete systems like organisms and integrated social groups may thus be considered as more or less “mindful” or “intelligent” creatures depending on how many of these capacities they have (Theiner and O’Connor 2010, 82-83). In a sense, all of the systems that posses one or more of these capacities may be regarded as cognitive agents although not all of them are purposeful (or intentional) agents. Furthermore, a big tent approach does not deny that there are other cognitive capacities in addition to those mentioned in the list.

The other perspective I would like to introduce here is the distributed approach to human cognition. The basic idea of this approach is that many cognitive processes in our everyday life transcend the boundaries of our skulls and skins since they are distributed across our environmental structures, information processing technologies, and the other people with whom we interact (e.g., Clark 1997; Hutchins 1995). The point is not just that we use different kinds of artifacts and technologies as aids in our cognitive processes as well as rely on the information that we obtain from other people. The distributed approach to human cognition (and the extended mind hypothesis) rather emphasizes that many of the cognitive artifacts, technologies, and organized groups profoundly alter the cognitive tasks that we face as individuals as well as transform our cognitive capacities both as technologically extended systems and as organized groups (Hutchins 1995; also Clark 1997; Kaidesoja 2012; Theiner and O’Connor 2010). However, this view does not amount to a denial of the central role of embodied and embedded human beings in all human cognition.

An important implication of the above perspectives is that they enable one to ascribe emergent cognitive capacities to social groups and to study the underlying mechanisms of these capacities empirically (e.g., Hutchins 1995; Theiner and O’Connor 2010). This nevertheless requires that we reconsider our received concept of cognition that ties all cognitive capacities to individual organisms (e.g., human beings), since groups obviously lack system-level consciousness or brains as distinct from those of their individual members. Proponents of distributed cognition thus hold that a kind of “big tent” approach to human cognition described above is needed, since it enables us to apply and adapt the methods and concepts of cognitive science to a broader range of cognitive processes taking place in our everyday life as well as to build bridges between cognitive and social sciences. So it is expected that this approach brings explanatory benefits when compared with the individualist and “brainboud” accounts of human cognition, including group cognition (e.g., Clark 1997; Hutchins 1995; Theiner and O’Connor 2010).

Now, drawing on organization studies (e.g., Scott and Davis 2003), I suggest that formal organizations (in short, organizations) can be understood as social groups that are designed to accomplish some (more or less clearly specified) goal or goals, and whose activities are planned, administrated, and managed by their members (or some subgroup of their members such as managers). Examples of organizations include schools, business firms, universities, hospitals, political parties, and governments. From the distributed cognition perspective, relatively enduring organizations can be considered as distributed cognitive systems that are composed of their individual members (performing more or less strictly defined and complementary roles) as well as the cognitive artifacts (e.g., strategies, policies, budgets, written rules, models, and ICT tools) and environmental structures (e.g., the spatial organization of offices, legislation that is relevant to the organization) they employ in materially and socially distributed cognitive processes (Hutchins 1995; Kaidesoja 2012; Secchi 2011). Drawing on these ideas, I suggest that at least some organizations (e.g., enduring business firms) are able to

adapt to (a certain range of) changes in their environment (e.g., market fluctuations),

Insofar as an organization possesses these five cognitive capacities listed by Theiner and O’Connor (2010, 82-83), I would say that it can be considered as a purposeful cognitive agent (with emergent cognitive capacities). It is nevertheless an empirical question whether, or to what extent, an organization of interest can be regarded as a cognitive agent.

Although the space limitations do not allow me to argue this point in detail, I propose that the distributed cognition approach could be useful in empirical studies on the emergent cognitive capacities (or capabilities) of organizations and their underlying mechanisms. This perspective suggests, for example, that the cognitive division of labor between members of an organization as well as their uses of cognitive artifacts and technologies in action coordination are crucially important for the emergent capacities of this kind. It does not imply, however, that every organization automatically outperforms individuals in every cognitive task, since it is clear that some cognitive processes (e.g., bandwagon effects) in organizations are counterproductive with respect to efficient functioning of an organization (e.g., Secchi 2011). Nevertheless, some of the underlying mechanisms of the emergent capacities of organizations have already been studied in organization research. Examples include studies on collective representations, collective decision-making, distributed/transactive memory, organizational learning, and problem solving in organizations (for an extensive review of organization studies, see Scott and Davis 2003; for organizational learning and memory, see Argote 2011; Austin 2003; Hutchins 1995; for decision making in organizations, see Secchi 2011). The perspective of distributed cognition has also been explicitly used in studies of this kind (e.g., Hutchins 1995; Secchi 2011).

6. Conclusion

The above arguments suggest that, insofar as they take the mechanism approach to explanation seriously, analytical sociologists have good reasons to overcome the position of IMF. They can do so by taking meso- and macro-level social mechanisms into account when studying large-scale social phenomena. I believe that this is required if their hope of providing “the central template for a renewed sociology for the twenty-first century” is to be realized. It was suggested, too, that historical macro-sociology, organization studies, and distributed cognition perspective provide fruitful insights that can be used to broaden the perspective of analytical sociology.

Some readers may still wonder whether the ontological position outlined above is compatible with the view that that all of the emergent causal powers of social entities supervene on the properties and relations of individuals (cf. Hedström 2005, 73-74). Although I have my doubts about the usefulness of the concept of supervenience in the context of social ontology, I will suggest here only that, strictly speaking, the supervenience base of the emergent powers of collective agents does not solely consist of properties of individuals and their relations. This is because in addition to interlinked individuals, also cognitive artifacts and technologies that are used by them in processing, storing, and transmitting information are constitutive of some of the emergent powers of collective agents.