Haohao Lei

Introduction

The causal mechanism is one of the key concepts in the field of analytical sociology. Mechanism-based explanations are becoming increasingly popular in the field of social science, history, and even biology. This essay will briefly discuss why mechanisms play an important role in sociological explanations. 

In the first part of the essay, I will try to define what a mechanism is. Then I will talk about why a causal mechanism is important. After discussing the mechanisms as concepts, I will use two specific mechanisms, namely signalling and diffusion, to explain why a social mechanism is essential to sociological explanation. Finally, I will talk about some of my concerns regarding the use of a mechanism-based perspective in sociological explanation and give a brief conclusion. 

Defining mechanism

As mentioned above, the term “mechanism” is widely used in various fields which investigate different types of subjects. For example, biology tends to investigate highly integrated systems, whereas evolutionary biology and social sciences tend to investigate dispersed systems. As a result, various definitions of “mechanism” are provided by different scholars in different fields. Thus, it is very difficult to provide a perfect definition of this concept. 

More importantly, as suggested by Hedström and Ylikoski (2010), even definitions of mechanisms by scholars within the same field are confusing. For example, Jon Elster, one of the most famous advocates of mechanism-based explanations, has two different definitions of a mechanism. The earlier version of the definition suggested that a mechanism is basically the “continuous and contiguous chain of causal or intentional links between the explanans and the explanandum” (Elster, 1989). This wording of the definition tends to wrongly imply that a mechanism does not need to be based on a causal relationship. On the other hand, Elster’s latest definition suggested the “mechanism is frequently occurring and easily recognizable causal patterns that are triggered under generally unknown conditions” (Elster, 1999). As Hedström and Ylikoski (2010) suggested, this definition is also problematic because there is no evidence that “frequently occurring” or “easily recognizable” should be key characteristics of mechanisms; on the contrary, some mechanisms are actually difficult to identify. Similarly, according to Little (1991, p. 15), a mechanism is “a series of events governed by law-like regularities that lead from the explanans to the explanandum.” Little’s definition of a mechanism is not accurate because it wrongly emphasises the characteristics of “law-like regularities.” As I will discuss in a later part of this essay, a mechanism should be considered an alternative way of providing explanation instead of being “law-like”. 

However, as suggested by Hedström and Ylikoski (2010), although the definition of a mechanism varies, all mechanisms tend to share some key characteristics. First, a mechanism is identified by the kind of effect or phenomenon it produces. Second, a mechanism is the basic unit of a causal notion; it cannot be further reduced. Third, mechanisms tend to have structures which illustrate how the entities’ properties, activities, and relations interact with each other and lead to the social phenomena. Fourth, mechanisms are hierarchical. The characteristics of an entity at one level could be explained by lower-level mechanisms, whereas a physical process might be the most fundamental mechanism, and there are no lower-level mechanisms anymore. 

Meanwhile, in practice, the word “mechanism” is also used to refer to slightly different notions. Sometimes it is used to refer to a causal process and sometimes to the essential elements of the causal process. However, this is a linguistic problem rather than a theoretical problem. 

Personally, I believe the definition of “mechanism” is similar to US Supreme Court Judge Potter Stewart’s definition of “obscenity.” As Judge Stewart observed, it is extremely difficult to provide a definition of obscenity, but “I know it when I see it”. Similarly, although it is difficult to provide a perfect definition of mechanisms, mechanisms are easy to identify with specific examples. 

The mechanism’s role in sociological explanation

In this part of the essay, I mainly will highlight the mechanism’s role in sociological explanation from three aspects. First, I will discuss why a mechanism-based explanation could act as a substitute to a law-based explanation and why a mechanism-based explanation is more suitable for explaining social phenomena. Second, I will briefly discuss how the development of a mechanism-based explanation is closely related to the counterfactual approach and the causal revolution in social science. Third, I will discuss how a mechanism could be used to link micro level and macro level social phenomena and how it is related to methodological individualism or structural individualism. Finally, I will briefly discuss how the mechanism-based explanation is related to the middle-range theory.

Law-based explanations are widely used in the field of physics and have been considered the most “scientific” method of explanation for a long time. However, the law-based explanation is difficult to identify outside the field of physics, especially in the recently rapidly developing scientific fields, such as biology and social science. As Cummins (2000) suggested, in social science, it is difficult to find any observable empirical regularities that support a law-based explanation. The mechanism-based explanation, however, is an alternative way of providing a scientific explanation. In addition to the fact that various mechanisms could be identified outside the field of physics, the mechanism-based explanation can also clearly show the asymmetry of explanatory relations. More specifically, a mechanism-based explanation could show the direction of the causality, whereas a law-based explanation fails to indicate which factor is the cause and which is the consequence. For example, according to Newton’s third law, not only can the action explain the reaction, but the reaction can also explain the action. This law does not allow the researcher to identify the direction of the causality. This is not a problem for the mechanism-based explanation as direction is a key characteristic of a mechanism. 

The mechanism-based explanation is also closely related to the increasingly important role of causality in social science. As suggested by many researchers, the mechanism-based explanation emphasises opening the black boxes and directly observing the “causal cogs and wheels” of the effect (Hedström and Ylikoski, 2010). The counterfactual approach of causal inference is particularly important in this case because the counterfactual approach of causality allows for the identification of the causation of a particular pathway without affecting any other part of the causal structure. Furthermore, the mechanism-based explanation is a particularly important tool in identifying the true causal relationships in correlations, especially in non-randomly controlled settings. As suggested by Hedström (2005), an existing correlation and plausible mechanism between A and B could support the causal relationship between A and B; more importantly, missing a reasonable mechanism linking A to B provides a good reason for rejecting the direct causal relationship between A and B even if there is a strong correlation between A and B. In the latter case, a more sophisticated causal diagram is required to explain the correlation between A and B. Therefore, a mechanism could refine the sociological explanation with more details and help in the search for more fundamental explanations. 

Mechanism-based explanation could also link micro and macro level social phenomena into a single explanatory framework. In most mechanism-based sociological explanations, the basic unit of explanation is the human agent. The mechanism-based explanation focuses on the actions and interactions of human agents and how these actions and interactions produce both micro and macro level social phenomena. According to the mechanism perspective, using one macro property to explain another macro property is not satisfactory. To explain the changes in macro level social phenomena, one needs to “open the black box” and investigate the mechanism which generates the macro level changes. As suggested by the famous Coleman’s boat illustrated in Figure one, the simple macro level association needs to be reduced to a chain of situational, action-formation, and transformational mechanisms. The mechanism-based explanation’s emphasis on the facts that the individual is the basic unit of analysis and that the macro level social phenomena are generated by individual-level mechanisms are closely related to the concept of methodological individualism, which suggested that social norms, traditions, and social institutions cannot serve as “bottom line” explanations. However, some forms of structural mechanisms are also involved in the mechanism-based social explanation, especially for transformational mechanisms, because the aggregations of individual actions are often closely related to structural arrangements beyond individuals’ atomistic attributes. 

Furthermore, the mechanism-based explanation is also closely related to the innovation of how people organize knowledge and what the desired type of theory is in social science. For a long time, researchers believed that scientific knowledge should be aggregated into a grand unified theory which aimed to explain everything and encompass all sociological knowledge. Examples of this kind of grand theory include Parsons’s theory of the social system and Marx’s historical materialism. However, forming such a grand unified theory is almost impossible. As Merton (1968) complained, the sociological hypotheses derived from the grand theory are scattered and unimpressive; grand theory offers too many concepts but only a few confirmed theories. The mechanism-based explanation, on the other hand, takes an alternative approach. It aims to provide key causal pathways between entities without describing all details. Meanwhile, mechanisms can also aggregate with each other to from a set of causal pathways. According to the mechanism-based view, mechanisms are not limited to particular cases; instead, social mechanisms could be shared between different sociology sub-fields and adapted to different situations. This explanatory perspective is closely related to middle-range theory, which aimed to explain certain kinds of social phenomena but not all of them. Middle-range theory depends on identifying causal mechanisms. Furthermore, middle-range theory, similarly to the mechanism-based explanation discussed above, emphasises the important role of micro and macro linkages. As suggested by Merton (1967), mechanisms act as the fundamental building blocks of middle-range theories. 

Figure one: Coleman’s boat (Hedström and Ylikoski, 2010)

Signalling as a mechanism

In addition to a theoretical discussion about the mechanism-based explanation, one can also focus on some specific examples. I will use two examples to illustrate how the mechanism-based explanation works in real life.

Signalling theory is a typical example of the mechanism-based explanation. According to signalling theory, individuals use signals to communicate with each other. However, some signals can be easily mimicked and used to send fake information, especially when the signaller can benefit from doing so. Therefore, most reliable information contains a cost. If the cost of the mimic is higher than the benefits, the information can be considered reliable. A good signal is that which only those with the truly desired characteristics can send and is very costly to mimics (Gambetta, 2011). 

In this case, I will focus mainly on using signalling theory to explain violent behaviours among criminals. Violence is normally considered one of the most irrational behaviours. However, sociologists, such as Gambetta (2011), have suggested that violence can be explained with the idea of signalling theory. 

As suggested by Gambetta (2011), the idea of signalling theory is widely used to explain the behaviours of criminals due to the unique characteristics of the underworld. Unlike people in normal society, criminals face severe constraints in communication because they need to conceal their messages from the police. Therefore, it is difficult to develop a reliable central institute responsible for communication. Meanwhile, because criminals are not regulated by law, most live in a low-trust environment. It is difficult to verify the information they receive. As a result, criminals need to develop a signalling system that can help them exchange relatively reliable information. Violence is one of the most common signals criminals use. Gambetta (2011) also suggested that violence and communication are closely related to each other. If an individual is famous for being violent, they are less likely to commit actual violent behaviours to prove their reputation. This reputation could minimize the amount of violent behaviours in the future and therefore minimize the cost due to the risk caused by the violence. However, an individual needs to commit actual violent behaviours to establish this reputation. Violence serves as a signal of an individual’s characteristic of a willingness to commit violent behaviour. The high cost of the violence increases the reliability of the signal. Unlike simply claiming oneself to be “courageous and tough,” conducting actual violent behaviour is more costly to mimic; therefore, more reliable information about one’s level of “toughness” can be sent by conducting violence.

This sociological explanation is a typical middle-range theory which aims to explain a typical kind of violence based on a causal mechanism. In contrast, Collins’s (2009) theory on violence aimed to explain all kinds of violence. Collins’s (2009) theory suggested that most micro-violence is largely influenced by emotion and dominated by tension and fear. Meanwhile, other relevant emotions such as anger also leads to incompetent violence because they tend to influence people’s estimations of causal importance (Collins, 2009). Collins (2009) further claimed that violence is sometimes not related at all to an individual’s choice. Situational conditions, such as state breakdowns and violent crowds, may force nonviolent individuals to conduct violent behaviours. Compared with Collins’s (2009) theory, signalling theory clearly identifies the causal mechanism behind violent behaviour; instead, Collins’s (2009) theory focuses not on identifying the cause of violence but on identifying some of the common characteristics of violence. On the other hand, unlike Collins’s theory, signalling theory only can be used to explain the violence among criminals, and it is difficult to generalize to other kinds of violence, such as domestic violence. However, the mechanism demonstrated by the signalling theory could be used to explain phenomena in other fields, such as international relations, where countries need to communicate with each other in low-trust environments. 

Positive feedback as a mechanism

Positive feedback is another example of a sociological mechanism which clearly demonstrates how a mechanism links the micro and macro levels. In this case I will focus on the role of positive feedback in explaining the diffusion of a social movement. Biggs (2003) discussed the potential mechanism that illustrates how the spread of protest information could lead to actual protest behaviours. Positive feedback, as suggested by Biggs, plays an important role in the diffusion of collective action both within groups and between groups. The within-group diffusion process is considered interdependence by Biggs (2003), who suggested that knowing a larger number of people joining a protest will make an individual more likely to take part. First, when more people are participating in the protest, the hope of success increases. In addition, when there are more people, there is a lower risk of being caught by authorities. Finally, more people means that there is more moral obligation to join the protest due to the within-group social pressure. Biggs (2003) suggested that between-group diffusion happens through the process of inspiration, which shows that people are more likely to protest if they know others are also protesting. The information about other protesters can, first, provide an opportunity for the new group of people to consider the possibility of taking part in a protest. Then, knowing that other people believe their collective action will succeed increases the hope for success in a new group of people. Finally, if other people have succeeded, a new group of people will hope for a successful protest; however, if other people have failed, the new group of people will have less hope. Mechanisms of interdependence and inspiration, which are based on the rational action theory, can be used to explain not only how the diffusion of protest information is related to protest behaviours but also why protests’ participation and expectations shift rapidly. Furthermore, the mechanisms of interdependence and inspiration also connected the micro level individual action the with macro level social movement diffusion. The mechanisms suggested by Biggs (2003) are typical examples of the transformational mechanisms illustrated in Figure one. 

Discussion

Although the mechanism-based explanation has a lot of strength, as I discussed above, I still have some concerns and questions about it. 

First, unlike the law-based explanation, understanding the mechanism-based explanation contributes nothing to one’s ability for prediction. As a typical law-based explanation, Newton’s law can be used to predict many natural phenomena, such as the motion of objects. However, by understanding the positive feedback mechanism of social movement diffusion, for example, it is still impossible to predict the diffusion of a social movement; the only thing to be learned from the positive feedback mechanism is that protests’ participation and expectations shift rapidly, but it is impossible identify the specific shifting point. The separation of prediction and explanation is a key characteristic of the mechanism-based explanation. I am wondering to what extent an explanation is still needed if the explanation cannot be used to predict what one wants to know? What would be the new purpose of an explanation if the prediction and explanation are separate? 

Another concern related to the mechanism-based explanation is the extent to which social phenomena can be explained by simple mechanisms. There is no evidence to support the idea that all social phenomena can be explained by simple mechanisms. What if certain social phenomena are the result of complex mechanisms which are beyond humans’ cognitive abilities? What if people could not understand the “causal cogs and wheels” even if they opened the black box? Do they still need to investigate these phenomena? 

The final question is about the mechanism-based explanation and empirical evidence. To what extent can one make sure the mechanism they discover is the true mechanism? A similar result may arise from various mechanisms. What kind of empirical evidence is needed to identify what mechanism is the true mechanism of certain social phenomena? Furthermore, what kind of empirical evidence is needed to justify that mechanism is not just a fancy version of a social narrative?

Conclusion

In general, the word “mechanism” is difficult to define but easy to recognize in a specific example. The mechanism-based explanation could act as an alternative to the law-based explanation and is more suitable for explaining social phenomena. The mechanism-based explanation is also closely related to the counterfactual approach, causal revolution in social science. In addition, the mechanism-based explanation could be used to link micro level and macro level social phenomena, with positive feedback being a typical example. Meanwhile, the mechanism-based explanation is related to the middle-range theory as was shown with the case of signalling theory.

Reference

Collins, R. (2009). Micro and macro causes of violence. International Journal of Conflict and Violence (IJCV), 3(1), 9-22.

Elster, J. (1999). Alchemies of the Mind: Rationality and the Emotions. Cambridge University Press.

Elster, J. (2015). Explaining social behavior: More nuts and bolts for the social sciences. Cambridge University Press.

Gambetta, D. (2011). Codes of the Underworld. Princeton University Press.

Hedstrom, P. (2005). Dissecting the social: On the principles of analytical sociology. Cambridge University Press.

Hedström, P., & Ylikoski, P. (2010). Causal mechanisms in the social sciences. Annual review of sociology, 36, 49-67.

Little, D. (1991). Varieties of social explanation: An introduction to the philosophy of social science.

Merton, R. K. (1949). On sociological theories of the middle range. Pp. 39-72 in On Theoretical Sociology. New York: The Free Press.

Merton, R. K., & Merton, R. C. (1968). Social theory and social structure. Simon and Schuster.

Michael Biggs, ‘Positive Feedback in Collective Mobilization: The American Strike Wave of 1886’, Theory and Society, vol. 32, 2003, pp. 217–54. 6