Mechanisms and Mechanismic Explanations

2021-02-20 龍子湖讀書會

Mechanismsand Mechanismic Explanations

Craver and Tabery’s account on mechanism

I shall begin with the entry「Mechanisms in Science」 in SEP,written by CarlCraver and JamesTabery1,which does not address mechanisms in social science in general and sociology inparticular. However, it provides the necessary background for us to discussmechanisms in the context of social sciences or sociology. I go through itstext, omitting where justifiable to omit.

Craver and Tabery:

1. TheRise of the New Mechanism

Twentiethcentury philosophy of science was largely dominated by logical empiricism.More a framework for doing philosophy of science than anycoherentset of doctrines, logical empiricism addressed a range of issues inphilosophy of science through the lens of the logical and mathematicalstructures constitutive of scientific thought and practice (see the entry onlogical empiricism). Logical empiricism tended, by and large, to focus onabstract,epistemic features of science, with little attention to scientificpractice. Physics was the dominant exemplar.

Thenew mechanical philosophy emerged around the turn of the twenty-first centuryas a new framework for thinking about the philosophy of science. Thephilosophers who developed this framework were, by comparison with the logicalempiricists, practitioners as well of the history of science andtended, by and large, to focus on the biological, rather than physical,sciences. Many new mechanists developed their framework explicitly as asuccessor to logical empiricist treatments of causation, levels,explanation,lawsof nature, reduction, and discovery.

Aswith logical empiricism, the new mechanical philosophy is less a systematic andcoherent set of doctrines than it is an orientation to the subjectmatter of the philosophy of science. The approach emerged asphilosophers and historians of science began to break from the once-standardpractice of reconstructingscientific inference with the tools of logicand, instead, to embrace detailedinvestigation of actual episodes fromthe history of science. The main tenets of logical empiricism had been underintense criticism for decades, and a new era of historically informedphilosophy of science had taken hold through the works of, e.g., Kuhn (1962),Laudan (1977), and Lakatos (1977). To many such scholars raised in thispost-logical empiricist milieu, it appeared that much of the practice ofcontemporary science (both in the laboratory and in print) was driven by thesearch for mechanisms, that many of the grand achievements in the history ofscience were discoveries of mechanisms, and that more traditionalphilosophy of science, for whatever reason, had failed to appreciate thiscentral feature of the scientific worldview.

Aspectsof the new mechanical philosophy began to emerge in the late 1960s. Fodor(1968), for example, contrasted mechanistic explanations (dealing withparts and their law-like interactions)with functional explanations in psychology.Wimsatt (1972a, 1976), building on the work of Simon (1962) and Kaufman (1971),argued repeatedly that the abstract and idealized structures of logicalempiricism were ill-suited to understanding how scientists discover and explaincomplexsystems at multiple levels of organization. Cummins (1975) provided anaccount offunctionalanalyses, characterizing a function as a contribution a component partmakes to the overall capacity of some system that includes that component.Salmon (1984, 1989) argued that empiricist views of scientific explanation arefundamentally flawed because they neglect causal mechanisms.Cartwright (1989) argued that the logical empiricist conception of a law ofnature is, in fact, a philosophical fiction used to describe the search forcapacities and nomological machines.

Thesestrands began to coalesce into an over-arching perspective in the 1990s. Theearliest clear statement of the new mechanism was Bechtel and Richardson's(2010 [1993]) Discovering Complexity.They self-consciously put aside logical empiricist concerns with theoryreduction and focused instead on the process by which scientists discovermechanisms (see Section 6 below). Soon after, Glennan argued thatmechanisms are the secret connexion Hume sought between cause and effect(1996), a thesis related to and partly inspired by Cartwright's focus oncapacities and nomological machines (Glennan 1997). Likewise, Thagard'sHow Scientists Explain Disease centeredthe search for causes and mechanisms in medicine (Thagard 2000; see alsoSection 6 below). Machamer, Darden and Craver's 「Thinking about Mechanisms」(Machamer, Darden, and Craver 2000; familiarly known as 「MDC」) drew thesestrands together and became for many the lightening rod of the new mechanistperspective. MDC suggested that the philosophy of biology, and perhaps thephilosophy of science more generally, should be restructured around thefundamental idea that many scientists organize their work around the search formechanisms.

2. TheConcept of a Mechanism

Theterm 「mechanism」 emerged in the seventeenth century and derived from Greek andLatin terms for 「machine」 (Dijksterhuis 1961). Descartesunderstoodmechanics as the basic building block of the physical world; in Le Monde, he proposed to explain diversephenomenain the natural world (such as planetary motion, the tides, the motion of theblood, and the properties of light) in terms of the conservation of inertialmotion through contact action (see the entry on René Descartes). Subsequently,the idea of mechanism has been transformed many times to reflect an evolvingunderstanding of the basic causal forces in the world (besides conservedmotion): e.g., attraction and repulsion (du Bois Reymond),conservation of energy (Helmholtz), gravitational attraction (Newton) (Boas1952; Westfall 1971; see also entries on Hermann von Helmholtz andIsaacNewton). The concept of mechanism has had an almost separate evolutionin the history of the life sciences (Allen 2005; Des Chene 2001, 2005; Nicholson2012), at times eschewing the metaphysical austerity embraced by Descartes andmany early mechanists.

Thenew mechanists inherit the word 「mechanism」 from these antecedents, but, intheir effort to capture how the term is used in contemporary science, havedistanced themselves both from the idea that mechanisms are machines andespecially from the austere metaphysical world picture in which all real changeinvolves only one or a limited set of fundamental activities or forces (cf.Andersen 2014a, b).

Mechanists havegenerally eschewed the effort to spell out necessary and sufficient conditionsfor something to be a mechanism. Instead, they offerqualitative descriptions designedto capture the way scientists use the term and deploy the concept in theirexperimental and inferential practices.

Threecharacterizations are most commonly cited:

· MDC: 「Mechanisms are entities and activitiesorganizedsuch that they are productive of regular changes from start or set-up to finishor termination conditions」 (2000: 3).

· Glennan: 「A mechanism for a behavior is a complex systemthat produces that behavior by the interaction of a number of parts, where theinteraction between parts can be characterized by direct, invariant,change-relating generalizations」 (2002: S344).

· Bechtel and Abrahamsen: 「A mechanism is astructureperforming a function in virtue of its component parts,componentoperations, and their organization. The orchestrated functioning of themechanism is responsible for one or more phenomena」 (2005: 423).

Each of these characterizationscontains fourbasic features: (1) a phenomenon, (2)parts, (3) causings,and (4) organization.We consider each of these in detail below.

Auseful canonical visual representation of a mechanism underlying a phenomenonis shown in Figure 1 (from Craver 2007). At the top is thephenomenon,some systemS engaged in behavior Ψ. This is the behavior of themechanism as a whole. Beneath it are the parts (the x’s) and their activities(theψ’s) organized together.2The dotted horizontal lines3reflect the fact that the parts andactivities are contained within, are components of, the mechanism engaged inthis behavior. Thus represented, mechanisms are decompositional in the sensethat thebehavior of the system as a whole can be broken down into organizedinteractions among the activities of the parts.

Figure 1. A visual representationof a mechanism (adapted from Craver 2007).

Inthe early literature, these different characterizations were often treated ascompetitors. Tabery (2004) argued instead that they reflect different, andcomplementary, emphases and intellectual orientations. Many mechanists haveadopted this ecumenical4stance. For example, Illari and Williamson offer a 「consensusconcept」 of mechanism:

A mechanism for a phenomenonconsists of entities and activities organized in such a way that they areresponsible for the phenomenon. (2012: 120)

Likewise, Glennan refers to 「minimalmechanism」:

A mechanism for a phenomenonconsists of entities (or parts) whose activities and interactions are organizedin such a way that they produce the phenomenon. (Glennan forthcoming: Ch. 2)

Theseecumenicalcharacterizations each include the four basic elements and are designedto make the characterization more inclusive. MDC’s insistence on the regularityof mechanisms is abandoned, for example, to accommodate mechanisms that workonly once or that work irregularly (Skipper and Milstein 2005; Bogen 2005; seealso Section 2.1.2 below). Bechtel and Abrahamson’s emphasis on the 「functions」is abandoned to accommodate mechanisms that serve no end and to distancemechanism from this loaded term so often opposed to mechanism (although seeCraver 2001a; Garson 2013; Maley and Piccinini forthcoming; see also Section4.5 below).

Theseecumenical characterizations intentionally downplay the fact that the term「mechanism」 is used differently in different scientific and philosophicalcontexts (see Levy 2013 and Andersen 2014a,b for alternative overviews of thedifferences). Indeed, much of the progress in the early years involved learningto recognize the many ways that the term 「mechanism」 can be used and the manycommitments that can be undertaken in its name. (For still othercharacterizations of mechanism, see Woodward (2002), Fagan (2012), Nicholson(2012), and Garson (2013)). Taking these ecumenical views as a starting point,we now consider the four basic components: 1) the phenomenon, 2) parts, 3)causings, and 4) organization.

2.1 Phenomenon

Thephenomenon is the behavior of the mechanism as a whole. All mechanisms aremechanisms of some phenomenon (Kauffman 1971; Glennan 1996, 2002). Themechanism of protein synthesis synthesizes proteins. The mechanism of theaction potential generates action potentials. The boundaries of a mechanism—whatis in the mechanism and what is not—are fixed by reference to the phenomenonthat the mechanism explains. The components in a mechanism are components invirtue of being relevant to the phenomenon.

MDC (2000) describe mechanisms asworking from start- or set-up conditions to termination conditions. They insistthat it is impoverished to describe the phenomenon as an input-output relationbecause there are often many such inputs and outputs from a mechanism andbecause central features of a phenomenon might be neither inputs nor outputs(but rather details about how the phenomenon unfolds over time). Darden,appealing to the example of protein synthesis, often associates the phenomenonwith the end-state: the protein (Darden 2006). Craver (2007), following Cummins(1975) and Cartwright (1989), often speaks of the phenomenon roughly as acapacity or behavior of the mechanism as a whole.

2.1.1 Producing, Underlying, and Maintaining

[...]5

2.1.2 Regularity

[...]6

2.2 Parts

Mechanistshave struggled to find a concise way to express the idea of parthood requiredof the components in a mechanism. The project is to develop an account that isboth sufficiently permissive to include the paradigmatic mechanisms fromdiverse areas of science and yet not vacuous.

Formalmereologies are difficult to apply to the material parts of biologicalmechanisms. Axioms of mereology, such as reflexivity (everything is a part ofitself) and unrestricted composition (any two things form a whole) do not applyin standard biological uses of the 「part」 concept.

Glennan(1996) recognized the difficulty of defining parthood very early on. Hisproposal:

Theparts of mechanisms must have a kind of robustness and reality apart from theirplace within that mechanism. It should in principle be possible to take thepart out of the mechanism and consider its properties in another context.(1996: 53)

Yet even this is perhaps too strong,given that some parts of a mechanism might become unstable when removed fromtheir mechanistic context. Later, Glennan (2002: S345) says that the propertiesof a part must be stable in the absence of interventions, or that parts must bestable enough to be called objects. This notion is perhaps too strong toaccommodate the more ephemeral parts of some biochemical mechanisms or of themechanisms of natural selection (Skipper and Millstein 2005; but see Illari andWilliamson 2010).

2.3 Causings

Mechanistshave disagreed with one another about how to understand the cause in causalmechanism. New mechanists have in general been at pains both (1) to liberatethe relevant causal notion from any overly austere view that restrictscausation to only a small class of phenomena (such as collisions,attraction/repulsion, or energy conservation), and (2) to distance themselvesfrom the Humean, regularist conception of causation common among logicalempiricists (see also the entry on the the metaphysics of causation). Four waysof unpacking the cause in causal mechanism have been discussed: conservedquantity accounts, mechanistic accounts, activities accounts, andcounterfactual accounts. (It should be noted that some mechanists have evolvedin their thinking about causation.)

2.3.1 Conserved Quantity Accounts

[...]7

2.3.2 Mechanistic Accounts

[...]8

2.3.3 Activity-Based Accounts

[...]9

2.3.4 Counterfactual Accounts

[...]10

2.4 Organization

Thecharacteristic organization of mechanisms is itself the subject of considerablediscussion.

2.4.1 Organization and Aggregativity

[...] 11

2.4.2 Varieties of Organization

[...]12

2.4.3 Modularity

[...] 13

2.4.4 Jointness

[...] 14

2.4.5 Levels

[...] 15

2.4.6Stable and Ephemeral Mechanisms

Finally, mechanists have found it necessary todistinguish between stable mechanisms, which rely fundamentally upon the moreor less fixed arrangement of parts and activities, and ephemeral mechanisms,which involve a process evolving through time without fixed spatial andtemporal arrangement (Glennan 2009). The time-keeping mechanism in a clock, forexample, is a relatively stable assemblage of components in relatively fixedlocations that work the same way, with the same organizational features, each timeit works. Ephemeral mechanisms, in contrast, involve a muchlooser kind of organization: itemsstill interact in space and time, but they do not do so in virtue of robust,stable structures. Many chemical mechanisms in a cell are like that (Richardsonand Stephan 2007). Ephemeral mechanisms are surely a primary focus ofhistorical sciences, such as archaeology, history, and evolutionary biology(Glennan 2009).

2.5 WhatMechanisms Are Not and What Are Not Mechanisms

Theterm 「mechanism」 has been used in many different ways to express many differentideas. The new mechanists』 appropriation of the term is thus likely to causeunhelpful associations, and their liberalization of the term is likely to raiseworries that the notion of mechanism has thereby been trivialized (see, e.g.,Moss 2012 and Nicholson 2012). Here, we first distinguish the new mechanismfrom other doctrines with which it shares both name and family resemblance. Wethen discuss some things to which the new use of the term 「mechanism」 does notapply.

2.5.1 What Mechanisms Are Not

Newmechanists have explicitly eschewed the following associations with the term「mechanism」:

1.Mechanismsare not necessarily deterministic. Mechanisms might be stochastic if, forexample, they are composed of stochastic activities (Bogen 2005, 2008a), or, ina more mundane sense (i.e., one consistent with determinism), because it isalways possible for one or more factors to interfere with the working of amechanism; one of the parts might be broken, or an unexpected preventer mightinterfere with the operation of a mechanism. The truth or falsity ofdeterminism, and its relevance to understanding the special sciences, is anindependent issue from the question of whether something is a mechanism.

2.Appealto mechanisms is not necessarily reductionistic. Mechanisms are often describedas multi-level, with activities at different levels being equally essential tohow a mechanism works. Mechanistic explanations might look up, down, or arounddepending on the choice of an explanandum and the presuppositions of theexplanatory context (Bechtel 2009a). Mechanists might be reductionists oranti-reductionists. That said, many mechanists opt for some form of explanatoryanti-reductionism, emphasizing the importance of multilevel and upward-lookingexplanations, without rejecting the central ideas that motivate a broadlyphysicalist world-picture (e.g., McCauley and Bechtel 2001; Craver 2007). (Forfurther discussion, see Theurer 2013; see also Sections 3.1 and 5 and the entryon reductionism in biology.)

3.Notall mechanisms are machines. Machines are human-made contrivances with eachpart added and organized by a designer to perform a function; biological andsocial mechanisms, in contrast, are products of evolution, broadly construed(Darden 2006), and so display ornate forms of organization in comparison withcontrivances. One machine might contain multiple mechanisms (a car, forexample, has mechanisms for braking, propulsion, playing music and climatecontrol). Machines are also capable of being both active and passive (a stoppedclock is still a machine); mechanisms, in contrast, have a productive aspectand are always doing something.

4.Mechanismsare not necessarily sequential or linear. Mechanisms can have feedback loopsand cycles wherein the output of the mechanism or components in turn influencesthe input of the mechanism or components in a subsequent iteration (Bechtel2011). Also, the interactions among components in a mechanism need not bedescribable by a linear equation.

5.Mechanismsare not necessarily localizable (Bechtel and Richardson 2010 [1993]).Components of mechanisms might be widely distributed (as are many brainmechanisms) and might violate our intuitive or tutored sense of the boundariesof objects (as an action potential violates the cell boundary). The assumptionof localization is often an important heuristic in the search for mechanisms;however, this heuristic often must be abandoned as the mechanism's organizationreveals itself.

6.Mechanismsare not limited to push-pull dynamics. Descartes』 mechanism had this feature,but (as noted above) the new mechanism explicitly liberalizes the notion toaccount for other kinds of causing.

7.Mechanismsare not just fictions/metaphors. When a scientist says that there is amechanism that makes proteins in living organisms, she is not just using amachine metaphor; rather, she is saying that there are in fact parts andactivities organized in living organisms such that they produce proteins.

2.5.2 What Are Not Mechanisms

Onemight object that there’s nothing left of mechanism once it sheds thesehistorical associations. One might suspect that it has been trivialized (Dupré2013).

Theidea of mechanism is a central part of the explanatory ideal of understandingthe world by learning its causal structure. The history of science containsmany other conceptions of scientific explanation and understanding that are atodds with this commitment. Some have held that the world should be understoodin terms of divine motives. Some have held that natural phenomena should beunderstood teleologically. Others have been convinced that understanding thenatural world is nothing more than being able to predict its behavior.Commitment to mechanism as a framework concept is commitment to somethingdistinct from and, for many, exclusive of, these alternative conceptions. Ifthis appears trivial, rather than a central achievement in the history ofscience, it is because the mechanistic perspective now so thoroughly dominatesour scientific worldview.

Yetthere are many ways of organizing phenomena besides revealing mechanisms. Somescientists are concerned with physical structures and their spatial relationswithout regard to how they work: an anatomist might be interested in the spatialorganization of parts within the body with minimal interest in how those partsarticulate together to do something. Many scientists build predictive models ofsystems without any pretense that these models in fact reveal the causalstructures by which the systems work. Some scientists are concerned withtaxonomy, sorting like with like without regard to how the sorted items cameabout or how they work. Finally, in many areas of science, there is a widelyrecognized and practically significant distinction between knowing that C(e.g., smoking) is a cause of E (lung cancer) and knowing how C causes E. Thisis not so much an ontological difference as it is a difference in the grainwith which one seeks to understand a system’s causal structure. In short, thereare many framework concepts in science, and not all of them can be assimilatedto mechanisms.

Butwhat, the critic might push further, does not count as a mechanism? Here aresome contrast classes:

1.Entities(or objects) are not mechanisms. Mechanisms do things. If an object is notdoing anything (i.e., if there is no phenomenon), then it is not a mechanism.

2.Correlationsare not mechanisms. Mechanisms explain at least many correlations, and manycorrelations can be used to characterize causal or mechanistic relations, butcorrelations themselves are not mechanistic. The same can be said of meretemporal sequences of events.

3.Inferences,reasons, and arguments are not mechanisms. Though there are mechanisms ofinference and reasoning, what makes something an inference or a reason islogical relation and not (merely) a causal relationship between premise andconclusion.

4.Symmetriesare not mechanisms. Many kind of symmetry are of fundamental importance indifferent areas of physics (e.g., translational symmetries, rotationalsymmetries). These are features of physical systems that are highly generalfacts or assumptions, not mechanisms.

5.Fundamental laws and fundamentalcausal relations are not mechanisms. If a law or causal relation isfundamental, then (by definition) there is no mechanism for it.

6.Relations of logical andmathematical necessity are not mechanisms. Such truths hold in all possibleworlds and so do not depend for their truth on facts about the causal structureof this world.

This is not an exhaustive list ofnon-mechanisms or non-mechanistic framework concepts. Yet it demonstrates thateven the liberalized concept of mechanism is neither vacuous nor trivial.

Mechanismic Explanation in Craverand Tabery’s account, and Znepolski’s analysis of the fall of Berlin Wall as anexample

I shall take Glennan’s 「minimalmechanism」 as the definition of mechanism because it is most inclusive, andshall also take Figure 1 as its diagrammatic representation. We shall discussit with particular reference to the sociological theory. Also, we shall discussit as a continuation to the case of TheGreat Event, that is, what we have achieved so far in the last lecture for SE (Social Explanation),HE (HistoricalExplanation) and CE (Cultural Explanation). Thesemiotic system is reproduced here.

Note: The link between 「actors」 and 「grounds for interaction」 is omitted. Thelink between 「actors」 and 「definitionsof the situation」 is implied by that between 「actors」 and 「interaction」.The interaction between actors is not an ontological entity but arealismic one because it persists in a string of present moments; it is in facta short segment of the course of action that already passed away. The causaldirection between structure-cum-society and definitions of the situation inH-D is indicated.

First, how should we understand whatis 「phenomenon」? I think it means something observable at least to theresearcher, who in our case is the sociological theorist. Taking Znepolski’stheory of event as an example, it is what the researcher can observe in thesequences of incidents, the chain of state-events and The Great Event, that is, what is data (lean fact) in these three, namely, definitions of the situation in every incident (= interaction) grounds for interaction related to everyincident, and The Great Event byitself. (Note: What can be observed in sequences of incidents and the chain ofstate-events has been observed in the incidents.)

Now if the researcher takes The Great Event (which is in thephenomenon) as the explanandum, then the parts in the supporting mechanismshould be the historical incidents in HE (HistoricalExplanation). (I do not use 「producing」, 「underlying」, and 「maintaining」 (seesubsection 2.1.1); instead I use 「supporting」.) Why the sequences of incidentsin SE (Social Explanation) are excluded from the mechanism? Myreason is that the researcher will surely select historical actorsfrom the actorsin OSR (Ontic Structural Realism, actually some ontologicalpresuppositions) with a narrative-wise parsimonious but adequate organizationof historical incidents in his mind. Again, these historical incidents (in theform of sequences) are organized into historical narratives within thepolyphonic text, that is, within the chain of state-events. (《史記》〈項羽本紀〉is an example.) Thus, the polyphonic text—as intertwined sequences ofhistorical incidents—is the mechanism supporting the phenomenon which includes The Great Event (the explanandum).

Now we have ascertained the organization ofthe mechanism as 「historical incidents—historical narratives—polyphonic text」,then historical incidents are its parts. Since the researcher selects historicalactors from the actors,every historical incident is an incident. What then is an incident? Let us goback to Figure 1. xiψ-ing, where i = 1, 2, 3, ..., is the ithincident, which sits in the semiotic system as follows:

CE grounds for interaction→ interaction = (structure-cum-society ← definitions of the situation) H-D

SE sequences of incidents ← incident

(Note: I have dropped the subscript i for the sake ofsimplicity.) The interaction, that is, xiψ-ing, is a short segment of the course ofaction that already passed away, cut out according to the sub-interactional definitions of the situation by theresearcher. It is the outcome of a participating actor’s interpretation (thatis, narrative and strategy, which can be iterative and is usually repetitive)in that segment, and as such it is usually the situation（處境）in which the actor finds himself after theinteraction. If he widens his vision, the situation can be the human condition（人之景況）or the social world（社會世界）, which are respectively the focus of Hannah Arendtin The Human Condition and that ofAlfred Schutz in The Phenomenology of theSocial World.

Suppose that this actorgives his definition of the situation(a famous term coined by W. I. Thomas) he is in, and the other actor(for the sake of simplicity, let us say there is only one) who has been ininteraction with him also gives his. (Note: Only the definition of the situation can be known to the researcher, not thesituation itself.) In this way, data= definitions of the situation andthe hypothesis to be tested with the databy the researcher will be about the structure-cum-society. (Note:Society-cum-society differs from the hypothesis about it in the same manner asthe situation is different from its definition.) Denote data and hypothesis by Dand H respectively. Then theresearcher’s analysis is the Bayesian representation: Pr(D|H), that is, the conditional probability of D (now is taken as the variable or variables) given H (now is takenas the parameter or parameters).

More specifically, the hypothesis His about the structure-cum-society against which D is given rise to by the actors in interaction. Inother words, D is given rise to bythe agencies (actors) against an otherness (structure-cum-society,presumably common to the agencies unless the researcher thinks otherwise).Structure-cum-society (as otherness) is brought along by actors(as agencies) via action into interaction. This is made possible by theontological presupposition that otherness is a non-subsistence by itself aswell as a supervenience with respect to the agencies (actors). It shouldbe noticed that interaction is now an otherness (a fat fact) in the eyes of theresearcher (he himself as an agency). I summarize the connections between thesignifiers in the semiotic system, the logical status of each signifier in H-D (Hypothetico-Deductivism), the signified correspondingto it in the sociological theory and who decides these particulars, as follows:

It should be pointed out that theconventional structure-versus-agency divide is replaced in this formulationwith the division of labour between the researcher and the actor,subsumed under By whom?. If the actor also gives adescription of structure-cum-society, it will be subsumed under The signified as a definition of the human condition orsocial world (the signified). In this way, the description ofstructure-cum-society by the researcher is a hypothesis tailored out from thesociological theory he espouses, that is, H under Logical status.

Furthermore, I incline to think thatthe problem of structure and agency in sociology is probably a result of aterrible oversight on the part the sociological theorist of the following twoall-important theoretical or methodological necessities:

(1) Differentiation of the situation (orhuman condition or social world) from its definition.

(2) Division of labour between theresearcher and the actor.

We can now consider grounds for interaction. Back to Figure1. What is that SΨ-ing in Phenomenon? If we stipulate that only data should be counted into thephenomenon, then we have the following:

(1) Definitionsof the situation (as data D) andstructure-cum-society (as hypothesis H) in H-D are respectivelythe 「x1」 part and the 「ψ-ing」 part of x1ψ-ing. In this way, interaction equals to x1ψ-ing,and it is in the mechanism. (Remember: We are concerned with historicalincidents only. We may call them 「historical interactions」.)

(2) Groundsfor interaction (as data not in D)in CE is the 「x2」 part of x2(x1ψ-ing)-ing,which is an expanded mechanism to include x2. To emphasize thestaying effect (a second-order effect, no doubt) of structure-cum-society, werewritten it x2(x1ψ-ing)ψ–ing, which can be simplifiedwith much loss of clarity as x2(x1ψ)ψ–ing.

(3) The Great Event, which is not in the mechanism, but in the phenomenon.

(4) Structure-cum-society (hypothesis Hin H-D), that is, the 「ψ)ψ)-ing」 part of x2(x1ψ)ψ-ing, is not data, and hence not in the phenomenon. Needless to say, theremaining 「x2(x1」 part is data. The Bayesian representation Pr(D|H) (whichis theresearcher’s analysis) makes sense only when the two parts are side-by-sidetogether, that is, as x2(x1ψ)ψ-ing.

What then is SΨ-ing in Phenomenon?(S is the abbreviation for the term 「System」 in Craver and Tabery’s account.) Ithink it can be understood by applying the operation called concatenation16(I use the symbol ※) in this way:

SΨ-ing → S※Ψ-ing = (x2(x1ψ)ψ)-ing = (x2(x1※ψ)ψ)-ing

Look at S※Ψ-ing. In it, S = (x2(x1 , and Ψ-ing = ψ)ψ)-ing. S is thus a two-level systemwith x1 and x2 respectively on the sub-interactional andthe interactional levels, and ψ-ing and Ψ-ing also. But in SΨ-ing, the part 「Ψ-ing = ψ-ing)ψ-ing)」 does not hold because theright-hand side is not in Phenomenon. As a result, SΨ-ing in Phenomenon does not equalto S※Ψ-ing in Mechanism.17They are cousins only: one withconcatenation and the other not. I suggest that only S※Ψ-ing be used. (Note: I refrain fromusing the adjective 「mechanistic」 in order to emphasize the focus on mechanismas realism in this kind of explanation.)

As a reminder, x2 (grounds for interaction, also some reference positions selectedfrom the S-B network of speech) is the work of the actor just as x1(definitions of the situation, somereference positions selected from the S-B network of speech) is, whereas ψ-ing(a description of structure-cum-society) is that of the researcher.

Now we can look at the HE, which is organizedas 「historical incidents—historical narratives—polyphonic text = chain ofstate-events → TheGreat Event」. We know now that a historical incident is an S※Ψ-ing (restricted to historicalincidents) in the mechanism. Historical narratives are in the first placesequences of historical incidents, that is,

sequence 1 = S11※Ψ11-ing―S12※Ψ12-ing― ...―S1n※Ψ1n-ing―...

sequence 2 = S21※Ψ11-ing―S22※Ψ22-ing― ...―S2n※Ψ2n-ing―...

...

sequence m = Sm1※Ψm1-ing―S12※Ψm2-ing― ...―S1n※Ψmn-ing―...

...

Needless to say, these sequences may cross one another.

The researcher makes some historicalnarratives, each of which summarizes some necessary details and salient featuresof the sequences (m = 1, 2, ...). These narratives are no longer simplesequences of historical incidents (which are backed up respectively by H-D at the sub-interactional and by CEand SE at the interactional levels) but coherent storiesafforded by the researcher’s narration. Finally, he synthesizes these storiesinto a polyphonic text. All the way, he is guided by the sociological theory heespouses.

Thus seen, the explanation for The Great Event requires HE, SE, CE and H-Dtogether and in that order (because explanation startsbackward from the explanandum, which in this case is The Great Event). We can call it ME(MechanismicExplanation) for The Great Event, anddenote the connections in it as:

ME = ◇HE―SE―CE―H-D

Note: The symbol ◇read 「Explanandum is inside」.

The Great Event). We can call it ME (MechanismicExplanation) for The Great Event, anddenote the connections in it as:

ME = ◇HE―SE―CE―H-D

Note: The symbol ◇ read 「Explanandum is inside」.

But it is also clear that ◇H-D, ◇H-D―CE and ◇SE―H-D―CE are also ME by themselvesbecause the explanandum for both is definitionsof the situation, which is in the phenomenon concerned.18 CE by itself is however not ME because itsstipulated explanandum is interaction, which is not in the phenomenon. We mayre-stipulate that so long as the phenomenon concerned is not empty theexplanandum of ME needs not be in it. But this re-stipulation makes theidea ofME vacuous as it would include almost all if not all kindsof explanation into it.

Now turn to Causings in Figure 1. What are thecausings in the case of ME = ◇HE? The first thingthat comes to mind is action. Peter Hedstrom in Dissecting the Social 19 quotes Karl Popper:

The types ofmechanisms we are looking for are those concerned with causes and consequencesof individual actions, because, as Popper (1994) 20expressed it, actions are the 『animatingprinciples』 of the social. (P.34)

I agree that action in OSR is Causings.Action in the theoretical sociology I propose (see Appendix) is inseparablybound to the actor in a most ontologically fundamental way:

The body [of the actor] is in [his] action,[his] action is in the present moment, the present moment is in [his] body.

Action is outside of Mechanism in Figure 1.

In addition, I propose that structure-cum-society(as the signified), which is different from the hypothesis H (as the signifier)in Pr(D|H) in the same manner as thesituation (the signified) is different from its definition (the signifier), isa causing in the case of The Great Event. It is outside of Mechanism in Figure 1. Inpassing, I note that the whole OSR is outside ofMechanism and Phenomenon, including the actors. Exclusion of the actoris exactly what Niklas Luhmann does in his system theory, though somesociological colleagues think it idiosyncratic. I agree with him, and I evenexclude the whole OSR. My reason is simple: Thesociological theorist (yes, the sociological theorist conducts positiveinvestigations) works with ideas about men and things, not men and thingsthemselves. Paraphrasing W. I. Thomas’s famous invention, it is its definitionthat the sociological theorist works with, not the situation.Action is outside of Mechanism in Figure 1.

We turn to organization of mechanism. Organizationis too narrow a task for the researcher with respect to the liberty he enjoysin his work. He can surely organize the mechanism, and he can even define it.For example, in ME = ◇HE he takes asexplanandum The Great Event in thephenomenon concerned, and in ME = ◇H-D he can as well take as explanandum definitions of the situation in the phenomenon concerned. Theresearcher in fact has the liberty to take a single incident, a sequence ofincidents or any other sensible choice as a mechanism different from that in ME = ◇HE―SE―CE―H-D so long as he canfind his desired explanandum in the phenomenon concerned.

Back to subsection 2.5 in Craver and Tabery’saccount on what mechanisms are not and what are not mechanisms. Our discussionso far on mechanismic explanation complies with it almost without an exception.I shall discuss the realism claimed in a small paragraph there:

[...] When a scientist says that there is amechanism that makes proteins in living organisms, [...], she is saying thatthere are in fact parts and activities organized in living organisms such thatthey produce proteins.

Simply put, she believes that the mechanism is real. We may call it「mechanismic realism」（機制主義實在論）, upon which ME is built. The ME for The GreatEvent(explanandum), that is, ME = ◇HE―SE―CE―H-D, is an example showing that in offering her explanationa scientist has to claim the reality of all that is included in it—includingits constituents H-D, CE, SE and HE. It is a big claimbut probably unavoidable for her as a sociological theorist, since there arealways some fat facts (hypotheses supported by data) in the mechanism concerned. I personally think that it is anabsolute imperative of science.

In this way,a mechanism is a scientific representation (which is always a limited one) ofthe social world (where actors had or have lived, are living and willlive) under the scrutiny of the sociological theorist. He constructs it makinguse of the symbolic representations (social territory and/or symbolic universe)made by theactors.Social territory（社會領地）and symbolicuniverse（象徵全域）reside on the S-B network of speech (stock ofknowledge about the social world, according to Alfred Schutz) as reference positions.(I have said several times in these series of lectures that the S-B network ofspeech is accessible to all, including the researcher.)

In this light, the social world (Schutz), the humancondition (Arendt) or the situation (Thomas) is no more than OSR, which I have stipulated earlier in this lecture to beCausings, and is hence outside of Phenomenon and Mechanism (see Figure 1).Hereafter I shall call OSR 「Social World」, abbreviated as SW. Also hereafter, the mechanismic explanation in ourexample will be represented by ME= SW◇HE―SE―CE―H-D (it actually represents thesemiotic system of ME) in order to remind us that thesocial world is nevertheless the ultimate concern of sociology although thesociological theorist is never able to study it positively (hence it can onlybe ontological presuppositions in sociology, in fact regardless of whether thespeculative project or the scientific), and, saddest to say, he is more likelythan not to forget it.

It is nothard to see that mechanismic explanation is afforded by the distinction betweenMechanism and Phenomenon. The distinction looks similar to the Stoic onebetween Depth（深處）where the course of action inheres and Surface（表面）on which the sequence of events (as the traces left by the course of action)settles. How do 「Mechanism—Phenomenon」 the modern distinction differs from「Depth—Surface」 its ancient counterpart? The two do not differ in one aspect,that is, both exercise the same strict rule: Only data can be allowed in the former’s Phenomenon and the latter’sSurface. In another aspect they do differ: the modern distinction allows data enter Mechanism whereas its ancientcounterpart bars it from Depth. In other words, Mechanism is derives fromPhenomenon in the former while Surface derives from Depth in the latter.Perhaps it is a matter of whether ontology is absent or present—it is absent inthe former but present in the latter. We should discuss this absence ofontology in mechanismic explanation when we come to Roy Bhaskar’s theory ofcritical realism.

It should be noticed that the modern distinction isapt for use by the researcher while its ancient counterpart by the actor.「Phenomenon-Mechanism」 is a special design by and for the researcher. It isnevertheless a successful design for positive investigations because it makesguessing the mechanism (an image of the social world) that seems to fit thephenomenon (all data, nothing but data) a progressive jigsaw puzzle21 exercise, in which knowledge accumulates forunending improvement and perfection. Much more importantly, this effort can becarried on by generations of researchers, an inter-generational effort which isa key feature of successful research projects in natural sciences. Howeve, itis not often seen in the field of sociological theory.

Mechanismic explanation in Bhaskar’s criticalrealism, and Clemens’s study of women’s groups and transformation of U.S.politics, 1890-1920 as an example

Bhaskar’s critical realism so far has not gained abroad reception among the English-speaking sociological circle.22 My attention to it was first drawn by the highregard it enjoys among a small group of sociologists including Margaret Archerand Philip S. Gorski. Then I noticed two terms, namely, 「mechanism」 and「event」, which are keywords in our discussion about sociological explanation,appear in a table (reproduced below) quoted in Steinmetz’s article23published in 1998 from Bhaskar’s A Realist Theory of Science24first published in 1975.

Domain of theReal Domain of the Actual Domain of the Empirical

――――――――――――――――――――――――――――――――――――――

Mechanisms +

Events + +

Experiences + + +

――――――――――――――――――――――――――――――――――――――

Bhaskar justifies it as follows:

[...] It hasoften been contended a constant conjunction of events is insufficient, but ithas not so far been systematically argued that it is not necessary. This can,however, be shown by a transcendental argument from the nature of experimentalactivity.

It is acondition of the intelligibility of experimental activity that in an experimentthe experimenter is a causal agent of a sequence of events but not of thecausal law which the sequence of events enables him to identify.

This suggeststhat there is a[n] ontological distinction between scientific laws and patternsof events. Obviously this creates a prima facie problem for any theory ofscience[.] I think that it can be solved along the following lines: To ascribea law one needs a theory. For it is only if it is backed by a theory,containing a model or conception of a putative causal or explanatory 『link』,that a law can be distinguished from a purely accidental concomitance. [...]Now at the core of theory is a conception or picture of a natural mechanism orstructure at work. Under certain conditions some postulated mechanisms can cometo be established as real. And it is in the working of such mechanisms that theobjective basis of ascriptions of natural necessity lies. It is only if we makethe assumption of the real independence of such mechanisms from the events theygenerate that we are justified in assuming that they endure and go on acting intheir normal way outside the experimentally closed conditions that enable us toempirically identify them. [...] Moreover it is only because it must beassumed, if experimental activity is to be rendered intelligible, that naturalmechanisms endure and act outside the conditions that enable us to identifythem that the applicability of known laws in open systems, i.e. in systemswhere no constant conjunctions of events 25prevail, can be sustained. This has the corollarythat a constant conjunction of events cannot be necessary for the assumption ofthe efficacy of a law.

[...]Similarly it can be shown to be a condition of the intelligibility ofperception that events occur independently of experiences. [...] Thus I willargue that what I will the domains of the real, the actual and the empiricalare distinct.

Bhaskar proposes three domains, namely, the Real, the Actual and theEmpirical. I presume that they exist in the natural world. The experimenter ispresumably situated at the level of Experiences, experiencing the three domains in the natural worldreally, actually and empirically respectively. (He must be a naturalscientist.) In which domain can he collects data? For sure, he can do so in the Empirical. Canhe do so in the Actual? Presumably he cannot. The design and setup of theexperiment—a closed system—presumably belongs to the Actual. It is within thisclosed system that Mechanisms generate Events which—as a constantconjunction—thesetup of the experiment—a closed system—presumably belongs to the Actual.It is within this closed system that Mechanisms generate Events which—as aconstant conjunction—the experimenter can experience actually. But in the opensystem (that is, outside the particular conditions imposed by the design andsetup of the experiment), Mechanisms are no longer in constant conjunction andso are Events, and the natural scientist (who has shaken off his experimenterrole) experiences Mechanisms really—and presumably directly.

Using the jargon we ourselves use, if what theexperimenter experiences empirically is data,then what he experiences actually and really will be hypotheses respectivelyabout Events and Mechanisms. Needless to say, hypotheses come from the theoryhe espouses, or even the theory itself is a set of hypotheses. No doubt, theexperimenter is an agency; hence Events and Mechanisms are othernesses. Thereis an ontological opposition between Mechanisms and Events on one side andExperiences on the other. Thus we obtain a different reading fromSteinmetz’s.

If my reading is correct, then Bhaskar’s tableshould be modified and elaborated as follows:

I rearrange it into a semiotic system for Bhaskar’stheory of critical realism as follows:

Note: The terms Cosmology and Depthrealism are borrowed from Whitehead26and Steinmetz 27respectively. The link between Experiences and data is deleted since data as empirical othernesses is nowconnected to experimenter as agency.

I make the following observations or comments onthis semiotic system:

(1) Cosmology and Ontology are respectively metaphysical and ontologicalpresuppositions.I follow thephenomenological tradition to delimit ontological entities to subjectivity andobjectivity and their derivatives. Bhaskar, being in the analytical tradition,is unlikely to agree with my labeling.28The natural world is an ontological entity in his eyes.

(2) Bhaskar’s original theory provides Cosmology and Depth realism and Isupplement it with Experiment and Ontology. It is now a more or less full semiotic system ofa theory about the natural scientist, almost ready for use in some positiveinvestigations by the social scientist. But if the social scientist isinterested in the social world instead of the natural world, he will study thesocial actor instead of the natural scientist, he perhaps needs a theory ofsocial action, and hence this semiotic system may be useless to him. I shalldiscuss this later.

(3) In connection to (2), I draw yourattention to the fact that experimenter as agency belongs and only belongs toClosed System. It implies that Open System which is the no-man’s-land in thesocial world is not interesting at all to the social scientist who isinterested in the social world. In other words, the social world is made ofinnumerable Closed Systems (nations, states, institutions, organizations,schools, families, etc.) populated with innumerable social actors, connectedbetween and among themselves and overlapping with one and other.

(4) Bhaskar’s theoretical ambition lies in Cosmology and Depth realism, theparts of the semiotic system he provides. The crux of the matter is thedistinction between Mechanisms and Events in Depth realism, andsubsequently that between Open System and Closed System in Cosmology. Steinmetz, half-quoting Bhaskar, summarizes itin one phrase 「mechanisms combine to produce actual events conjecturally, thatis, in concert with other mechanisms」.29 I shall goback to Bhaskar’s own text and discuss it later.

(5) If one follows through the crux ofBhaskar’s matter, one will question: What exactly does he want to explain?

(6) In connection to (4), it should be notedthat hypothetico-deductivism slips in unnoticed since actual hypotheses and data are present in Experiment.

(7) Does Bhaskar use the term 「mechanism」 inthe way it is meant in Craver and Tabery’s account of mechanism?

(8) Do Bhaskar’s 「event」 and Znepolski’s「state-event」 mean the same thing?

Bhaskar:

[...] But theaim of science is the production of the knowledge of the mechanisms of theproduction of phenomena in nature that combine to generate the actual flux ofphenomena of the world. These mechanisms, which are the intransitive objects ofscientific enquiry, endure and act quite independently of men, The statementsthat describe their operations, which may be termed 『laws』, are not statementsabout experiences (empirical statements, properly so called) or statementsabout events. Rather they are statements about the ways things act in the world(that is, about forms of activity of the things of the world) and would act ina world without men, where there would be no experiences and few, if any,constant conjunctions of events. [...] 30

If mechanisms are intransitive objects ofscientific enquiry then events are their transitive counterparts. By「transitive」 it means being dependent of men (experimenters, naturalscientists). Elsewhere Bhaskar has defined a closed system as one in which aconstant conjunction of events obtains,31 thus the open system will be as he says herehaving 「few, if any, constant conjunctions of events」. Note that mechanismsgenerate events conjuncturally regardless of whether the system concerned isopen or closed; the difference is only that events in the open system areseldom in constant conjunctions (so Bhaskar believes) whereas those in theclosed one are.

It means that men (who are in the closed systemalready) intervene with the conjuncture of mechanisms (yes, mechanisms)concerned, though not always successfully. There must be unsuccessful cases inwhich events in the closed system are not in constant conjunction; alsofailures must be far more often in the social world (where social actors stay)than in the natural one (where natural scientists stay), and hence nations,states, institutions, organizations, schools, families, etc. (closed systems)in the social world must need constant maintenance (law and order, safety andsecurity, etc.). Steinmetz sums it up in one sentence 「Constant conjunctions[of events] have to be produced artificially.」32

Bhaskar’s conception of mechanism and event doesnot tally with Craver and Tabery’s account. First, he does not accept data (in Domain of the Empirical) intomechanism (in Domain of the Real). (Depth realism ishierarchical, a generative cascade from Mechanisms down to Events and furtherdown to Experiences.) Second, Craver and Tabery do not mention specifically theidea of event, and I assume that they will allow Znepolski’s state-events to beincluded in their conception of mechanism.the idea of event, and I assume that they willallow Znepolski’s state-events to be included in their conception of mechanism.

Bhaskar’s conjunction of events is not the same asZnepolski’s chain of state-events. The latter is a polyphonic text which iscomposed out of historical narratives, which in turn are composed out ofhistorical incidents related to historical actors. Furthermore, If Craver andTabery’s account is followed, Znepolski’s chain of state-events is not data and hence is not in theirphenomenon. But Bhaskar takes his conjunction of events as 「the actual flux ofphenomena of the world」, that is, it is in the domain of the actual, andtherefore is not data in that of theempirical.

Look at Znepolski’s mechanismic explanation of The Great Event, that is, ME = ◇HE―SE―CE―H-D. We take itformally, that is, disregarding its content, and see whether it can fitformally into the semiotic system of Bhaskar’s theory of critical realism. Forthe sake of simplicity, we drop SE and CE without much loss of generality, resulting in ME = ◇HE―H-D. Now, the pair of opposites 「data—actual hypotheses」 inExperiment is identical to H-D in ME. That is done. Now, can the other pair of opposites「actual hypotheses—real hypotheses」 in Experiment be similarif not equal to HE? No, it cannot because ◇ (Explanandum is inside)(remember: an explanandum must be datain Craver and Tabery’s account of mechanism) cannot be placed on either side ofit since neither actual hypotheses nor real hypotheses are data.

Thus it is clear that the problem is with theexplanandum: What should be taken as the explanandum in Bhaskar’s theory?Notice that data is in H-D. Alright, we can reverse the order of SE and H-D, thus obtain ME = ◇H-D―HE, which is still a mechanismic explanation, whoseexplanandum is data in the domain ofthe empirical. But it is no longer the same as Znepolski’s.

Since Bhaskar takes the conjunction of events(actual hypotheses) to be the phenomenon in his theory, he perhaps wants it tobe the explanandum for his mechanismic explanation. Putting aside for a whileCraver and Tabery’s version of phenomenon, we can perhaps rearrange Bhaskar’sversion of mechanismic explanation as follows:

Craver and Tabery’s definition of Phenomenon is thus relaxed from beinglean fact (data) to being fat fact(actual hypotheses—data), and at thesame time their definition of Mechanism istightened to admit only real othernesses (realhypotheses). Bhaskar’s version of mechanismic explanation is thus:

ME = NW◆H-D―TE

where NW and TE are respectivelyabbreviations of 「Natural World」 (surely no more the SW as in the ME for The GreatEvent) and 「Transcendental Explanation」 (since the explanans is nothing butthe real hypotheses). ◆ should read 「Explanandum is this」. Perhaps ◆ can onlybe parked against an explanation like H-D that contains somegrain of data (empirical othernesses)to gain its worth of explanation.

Bhaskar continues:

Movingtowards a conception of science as concerned essentially with possibilities,and only derivatively with actualities, much attention is given to the analysisof such concepts as tendencies and powers. Roughly the theory advanced here isthat statements of laws are tendency statements. Tendencies may be possessedunexercised, exercised unrealized, and realized unperceived (or undetected) bymen; they may also be transformed. Although the focus of this study is naturalscience, something is said about the social sciences and about thecharacteristic pattern of explanation in history 33

Bhaskar is correct in asserting that his criticalrealism is 「concerned essentially with possibilities」, since his mechanismicexplanation is a fact fattened twice—first by actual hypotheses and then byreal hypotheses. The real hypotheses (explanans) is quite remote from the leanfact (data), and hence ratherspeculative, or in his words 「concerned essentially with possibilities」.

What are tendencies and powers in the naturalworld? Bhaskar explains:

If theanalysis of causal laws (and generative mechanisms) is to be given by theconcept of things and not events [...], the consideration that they not onlypersist but are efficacious in open systems [...] entails that causal laws mustbe analyses as tendencies. For tendencies are powers which may be exercised withoutbeing fulfilled or actualized (as well as being fulfilled or actualizedunperceived by men). It is by reference not just to the enduring powers but theunrealized activities or unmanifest (or incompletely manifest) actions ofthings that the phenomena of the world are explained. It is the idea ofcontinuing activity as distinct from that of enduring power that the concept oftendency is designed to capture. In the concept of tendency, the concept ofpower is thus literally dynamized or set in motion.34

For the sake of easy reading, simply take 「laws andgenerative mechanisms」 as mechanisms in Bhaskar’s theory of critical realism.Also, a suspected slip of pen by Bhaskar should be corrected: Since activitiesare things in the domain of the real they cannot be 「unrealized」. Perhaps hemeans 「realized but unmanifest」. I take the liberty to replace it with「realized but incompletely manifest」.

He makes two claims:For the sakeof easy reading, simply take 「laws and generative mechanisms」 as mechanisms inBhaskar’s theory of critical realism. Also, a suspected slip of pen by Bhaskarshould be corrected: Since activities are things in the domain of the real theycannot be 「unrealized」. Perhaps he means 「realized but unmanifest」. I take theliberty to replace it with 「realized but incompletely manifest」.

He makes two claims:

(1) Things by themselves alone arepresupposed to be capable of continuing activities (or actions) in open systems(the no-man’s-land).

(2) The continuing activities of things arerealized but incompletely manifest by themselves in open systems but may beactualized by men (who can be natural scientists or social actors) in closedsystems (the men’s-land).

Clearly both claims are cosmologicalpresuppositions. As continuing realized but incompletely manifest activities ofthings, they are tendencies. As tendencies of things that may be actualized bymen, they are powers. 35

A classical example of what Bhaskar means by「tendencies」 is Newton’s Laws of Motion. They are powerful statements (realhypotheses, real othernesses in opposition to the experimenter as agency) whichtook more than a century to be 「transformed」, that is, to find a counter-example(data, empirical othernesses).Newton’s Laws (realized but incompletely manifest?) as tendencies reside as asingle reference position on the S-B network of speech (an open system, inBhaskar’s sense). But they can be taken out from it by men at any time to beused in his work, so they are 「powers」 in Bhaskar’s theory of critical realism.36

Bhaskar continues:

[...] ifscience is to be possible the world must consist of enduring and transfactuallyactive mechanisms; society must consist of an ensemble of powers irreducible tobut present only in the intentional actions of men; and men must be causalagents capable of acting self-consciously on the world, The do so in anendeavour to express to themselves in though the diverse and deeper structuresthat account in their complex manifold determinations for all the phenomena ofour world. 37

At this point Bhaskar moves from the natural world to the social world. Thework need be done is to connect the semiotic system of Bhaskar’s theory ofcritical realism and that of our theoretical sociology, as follows:

Note: → should read 「direction of flow」. Terms belonging to Bhaskar’stheory are italicized, and those appearing neither explicitly in Bhaskar’stheory nor our theoretical sociology but existing implicitly in both are shaded.Consult Appendix on the semiotic system of our theoretical sociology.

The two semiotic systems are articulated by meansof two terms, namely, 「actor」 and 「reference position」. The open system in Bhaskar’s theory is justareferenceposition on the S-B network of speech in our theoretical sociology as Ihave demonstrated minutes again. It is a kind we have not considered before,and I shall deal with it when opportunity arises. On the other hand, since theexperimenter is just a particular kindof actor,mainly as an agency in the closedsystem in Bhaskar’s theory giving rise to three particular kinds of othernesses(empirical othernesses, actual othernesses and real othernesses) which willbring some particularities into the sociological theory concerned but not ourtheoretical sociology. In this sense, our theoretical sociology in factrecruits that sociological theory like any other one. In other words, theconnection of Bhaskar’s theory of critical realism to our theoretical sociologyis just the recruitment of it as an exemplar. A passing remark: closed system is not areferenceposition.

In what follows, I shall give an example ofBhasker’s mechanismic explanation except that NW is replaced by SW, that is, in the form ME = SW◆H-D―TE. I have chosen forexample Elizabeth S. Clemens’s 「Organizational repertoires and institutionalchanges」38published in 1993.

Clemens gives the following summary of her papertowards its end:

Throughongoing processes of organizational innovation—the constant search forpolitical advantage or shared identity by trying something new, adopting somealternative model of organization—women's groups helped to create a new systemof political institutions. In the place of a political system in which votinghad been the central act and identity was grounded in the solidary networks ofcommunity and workplace, the beginning of the 20th century saw the rise of apolitical regime in which groups claiming to represent categories of personspresented specific demands to legislatures, using the leverage of publicopinion, lobbyists, and expertise rather than sheer numbers of votes. Althoughscholars may well differ over the conservative nature of these developments andtheir normative status, these changes stand as an important example of profoundinstitutional transformations stemming from regular—not revolutionary—politicalprocesses.their normative status, these changes stand as an important exampleof profound institutional transformations stemming from regular—notrevolutionary—political processes.

To understand how such change occurs, it isnecessary to abandon the assumptions of unilinear development and institutionalhomogeneity that have dominated theorizing about the relation of socialmovements to political institutions. Instead, recent developments inorganization theory and social movement studies point to the importance of amultiplicity of organizational models, a repertoire of organization.39 This variety forces qualification of the classicmodels of the relation between movements and political institutions. First,different models of movement organization (and differences in the identities ofthose organized) mean that some movements may be more susceptible to the logicsof incorporation that characterize a specific political regime. Second, inorder to circumvent the disadvantages imposed by a specific regime, movementgroups may import models of organization that are already culturally legitimatealthough not previously recognized as political. By using models oforganization that are simultaneously familiar and novel, social movement groupsmay bring about changes in the taken-for-granted rules about what politicalorganization is and what it is for. In the United States, the loosely knit「woman movement」 of the turn of the century provides an important example ofthis type of institutional change. Although the women’s parliaments and partiesdid demonstrate the isomorphism with existing political institutions predictedby the classic model of Michels, the majority of organizational activity bywomen’s groups involved a much more eclectic process of copying andtransforming multiple models of organization: 「Colleges and social scientistsand experts of various kinds can help us in thematters upon which we are working, but as to the ways of working wehave to blaze our own trail」 (Winter 1925, p. vi; emphases in original).Limited by their exclusion from the organization and practices of electoralpolitics, women’s groups were particularly motivated to discover or invent newchannels for their political activities. By drawing upon available alternativemodels of organization—business methods, state bureaucracy, and lobbies, alongwith models drawn from education and the professions—women’s groups helped topioneer a distinctively nonelectoral style of social politics. The success oftheir experiments was such that this new style was quickly appropriated byother political actors and the historical origins of this model of politicalorganization have been forgotten as the system of interest-group bargaining istaken as natural, indeed as constitutive of American politics (Moe 1980, p. 2).

Women’s groups were a source ofpolitical change because they were marginal to the existing electoral system,but not so marginal that they were ignored by other political actors. Togetherwith the assumption of organizational heterogeneity—the assumption that a repertoire of organization exists—thepresence of differences in political power is fundamental to this account ofinstitutional change. The potential of a challenging group to produce changesin existing institutions is a joint product of the incentives to innovateproduced by relative marginality andits visibility within the political arena, as well as the acceptability ofthose innovations to other political actors. Framed in these terms, thisprocess of institutional change is a recurring element in the political historyof this country. The civil rights movement, for example, was grounded in boththe decreasing political marginality and increasing organizational innovationof blacks in the United States. The Great Migration out of the South increasedblack voting strength just as the shift of blacks to the Democratic partycontributed to their leverage over the administration. Profiting from thisgrowing visibility, activists synthesized models of organization grounded inreligion with the strategies of nonviolent resistance and court-centeredcontestation to create a style of oppositional politics now. Migration out of the South increased black votingstrength just as the shift of blacks to the Democratic party contributed totheir leverage over the administration. Profiting from this growing visibility,activists synthesized models of organization grounded in religion with thestrategies of nonviolent resistance and court-centered contestation to create astyle of oppositional politics now shared by both the Left and the New Right(McAdam 1982). Similarly, the antiwar movement of the 1960s sought to exploitnew relations with the mass media and, thereby, helped to usher in a politicalsystem in which a central role is played by the access of challenging groups totelevision coverage and the ability of established elites to control the termsof that coverage (Gitlin 1980). Nor is this dynamic limited to the UnitedStates. Describing the rise of ecology parties in Western Europe, HerbertKitschelt (1989, p. 3) argues that 「Left-libertarians have engaged in protestmovements with loose alliances of federated, egalitarian organizations withlittle hierarchy or formalization of decision-making procedures. They haveattempted to build their parties in the same mode」 in the hope of ultimately「creating a more decentralized, libertarian and participatory society with lessemphasis on economic competition and growth.」 In those nations where it ismarginalized by the political entrenchment of social democratic parties, thelibertarian left has used the relatively high economic and educational capitalof its membership to make visible a logic of politics defined by therepresentation of alternative life-styles or values rather than by electoralcompetition and compromise.

This account of institutional change does notimply that challenging groups achieved all that they desired, that oppositionalintentions were not co-opted. Indeed, the enduring power of Michels’s analysisstems from the broad scope of the logics of incorporation in modern societies.But not all social groups and organizational forms are equally susceptible.Given the organizational heterogeneity of modern society and its consequencesfor the repertoires of organization that inform political life, the very processof challenging political institutions can change the rules of political action,if not necessarily the substance of political outcomes.

Look at ◆H-D the explanandum,that is, (Events—Experiences) = Phenomena, first. Let us gather whatare data, that is, empiricalothernesses found by Clemens the researcher, in her summary. I take definitions of the situation (or thehuman condition or the social world) as an example par excellence of empirical othernesses. Note that they areself-definitions, that is, definitions by actors themselves. A keyword (to be exact, a sign) in the textis recognized as an empirical otherness only if it is (or is believed to be)appointed by actors themselves. This recognition rule serves as the definingcharacteristic of empirical othernesses. I manage to gather the following fromClemens’s summary, shaded in the quotation concerned:

· 「[...] the majority of organizational activities by women’s groupsinvolved a much more eclectic process of copying and transforming multiple models oforganization: [...]」

· 「Instead, recent developments in organization theory and social movementstudies point to the importance of a multiplicity of organizational models, arepertoire of organization.」

· 「First, different models of movement organization (and differences in the identitiesof those organized) mean that some movements may be more susceptible to thelogics of incorporation that characterize a specific political regime.」

Notice that there is no definition of the situation or the human conditionor the social world. Its absence from the summary is perhaps due to itsrelative unimportance in historical institutionism （歷史制度主義）which is the school Clemens belongs. There is an oblique mention ofidentities (or more probably self-identities, I believe) of actors.

Next, I take structure-cum-society as an example par excellence of actual othernesses.The recognition rule is: a keyword is an actual otherness if—

(1) it is not an empirical otherness; and

(2) it is appointed by the researcher to afford adescription of structure-cum-society that is (or is believed to be) an acceptedhypothesis (actual othernesses) tested (whatever it may mean) with data (empirical othernesses).

I manage to gather the following:

· 「Through ongoing processes of organizational innovation—the constantsearch for political advantage or shared identity by trying something new,adopting some alternative model of organization—women’s groups helped to createa new system of political institutions.」

· 「In the place of a political system [...], the beginning of the 20thcentury saw the rise of a political regime in which groups claiming torepresent categories of persons presented specific demands to legislatures,...」

· 「To understand how such change occurs, it is necessary to abandon the assumptionsof unilinear development and institutional homogeneity that have dominatedtheorizing about the relation of social movements to politicalinstitutions.」

· 「Second, in order to circumvent the disadvantages imposed by a specificregime, movement groups may import models of organization that are already culturallylegitimate although not previously recognized as political.」 (Lui:cultural legitimacy)

· 「The success of their experiments was such that this new style was quickly appropriatedby other political actors [...]」 (Lui: appropriation of models)

· 「The potential of a challenging group to produce changes in existinginstitutions in a joint product of the incentives to innovate produced byrelative marginalityand its visibilitywithin the political arena, as well as theacceptability of thoseinnovations to other political actors. 「

· 「Profiting from this growing visibility, activists synthesized models oforganization grounded in religion with the strategies of nonviolentresistance and court-centered contestation to create a style ofoppositionalpolitics [...] Similarly, the antiwar movement of the 1960s sought toexploit new relations with the mass media [...] Despite the rise of ecologyparties in Western Europe, [...] 「egalitarian organizations with littlehierarchy or formalization of decision-making procedures [...]」 [...] the libertarianleft has used the relatively high economic and education capital of itsmembership [...]」 (Lui: egalitarianism, libertarianism). Lastly, an otherness that is neither empirical noractual is real. I manage to gather the following:

· 「First, different models of movement organization (and differences in theidentities of those organized) mean that some movements may be more susceptibleto the logicsof incorporation that characterize a specificpolitical regime.」

· 「To understand how such change occurs, it is necessary to abandon theassumptions of unilinear development and institutional homogeneitythat have dominated theorizing about the relation of social movements topolitical institutions.」

With these chosen keywords, I draw up the semioticsystem of Clemens’s summary, as follows:

where α is real othernesses, β1 and β2 parts ofactual othernesses, and γ empirical othernesses. Following Bhaskar’s criticalrealism, α is the mechanism, and β1, β2and γ thephenomena. The mechanism is supposed to be able to explain the phenomena.

Said earlier, empirical othernesses is appointed bythe actorswhereas actual othernesses by the researcher. In other words, the two arerespectively D the dataand H the actual hypotheses in Pr(D|H) which is the conditionalprobability representing H-D. Clearly there are many if nottoo many hypotheses to test withdata!

Perhaps it is not as bad as it seems because some of the actual othernessesmay have been also appropriated by the actors and become empirical ones. Thedividing line between the two kinds of othernesses is perhaps blurred. This is especially true if actors define the situation, the humancondition and the social world they are living in. (I have noted the absence ofsuch definitions in Clemens’s summary earlier.) I believe if they did, many ofthe actual or real othernesses would appear as empirical ones in the semioticsystem. I consider this a shortcoming, and believe that it probably weakens theargumentative power of historical institutionism. Notice that—

(1) β1 and γ (as signifiers) signifythe human condition or the social world (as the signified); and

(2) if γ signifies the structure (system,regime) in Znepolski’s theory of incident and event, then mass media, ecology,religion in β2 are society.

Thus structure-cum-society (which is the hypothesis H of H-D with definitions of the situation (or the humancondition or the social world) as datain Znepolski’s example) will emerge in Clemens’s summary. I may perhaps ventureto say that such definitions are sadly missing if not suppressed in it.

It should be pointed out that Clemens is veryunlikely to have adopted Bhaskar’s critical realism. One can only say thathistorical institutionism exhibited in her summary happens to fall intoBhaskar’s theory of mechanismic explanation, that is, ME = SW◆H-D―TE.40

Concluding remarks

The crux of the matter is this: What should becounted as phenomena? Clemens is likely to consider that her investigation isgrounded in what I call sociological theory (a social theory that under certaincondtion can sustain positive investigation by the researcher), and yet becauseof the thinness of data(empiricalothernesses) and the thickness of actual hypotheses (actual othernesses)— bothas phenomena—it seems to be too close to what I call social theory (a theorythat has not been positively investigated by the researcher). (see Appendix forthe classification of theories into social theory, sociological theory andtheoretical sociology.) Perhaps it is not a problem in the eyes of mostsociologists though I personally insist a strict distinction between socialtheory and its sociological counterpart.

On the other hand, Znepolski’s analysis (with myadding H-D and CE to it) complieswith the strict rule of natural science that only data should be counted as phenomena. We know now that Znepolskiachieves this by taking the fall of the Berlin Wall (surely a datum) as theexplanandum. Clemens can do the same because the women’s voting right has beenwon decades ago, and that winning of voting right can be taken as theexplanandum, and the rest will follow suit as in the Znepolski’s case. But itis a matter entirely left to Clemens to decide.

More generally speaking, historical institutionists should perhaps examinethe theory of mechanism they are using and start to think about this impendingdecision on which theory of mechanism should they follow.

1、Craver, Carl and Tabery, James, "Mechanisms inScience", The StanfordEncyclopedia of Philosophy (Winter2016 Edition), Edward N. Zalta (ed.), forthcoming URL =<http://plato.stanford.edu/archives/win2016/entries/science-mechanisms/>.

2、Some typos, corrected.

3、The「dotted horizontal lines」 is not found on Figure 1, probably a writing mistake.Figure 1 presented here is modified and redrawn.

4、Merrian-WebsterDictionary:

「FullDefinition of Ecumenical

「1: worldwide or general in extent, influence,or application

「2a: of, relating to, or representing the whole of a body of churches

「 b: promoting or tending toward worldwideChristian unity or cooperation」

5、Omitted:

「New mechanists speakvariously of the mechanism as producing, underlying, or maintaining thephenomenon (Craver and Darden 2013). The language of production is best appliedto mechanisms conceived as a causal sequence terminating in some end-product:as when a virus produces symptoms via a disease mechanism or an enzyme phosphorylatesa substrate. In such cases, the phenomenon might be an object (the productionof a protein), a state of affairs (being phosphorylated), or an activity orevent (such as digestion). For many physiological mechanisms, in contrast, itis more appropriate to say that the mechanism underlies the phenomenon. Themechanism of the action potential or of working memory, for example, underliesthe phenomenon, here characteristically understood as a capacity or behavior ofthe mechanism as a whole. Finally, a mechanism might maintain a phenomenon, aswhen homeostatic mechanisms hold body temperature within tightly circumscribedboundaries. In such cases, the phenomenon is a state of affairs, or perhaps arange of states of affairs, that is held in place by the mechanism. These waysof talking can in many cases be inter-translated (e.g., the product isproduced, the production has an underlying mechanism, and the state of affairsis maintained by an underlying mechanism). Yet clearly confusion can arise frommixing these ways of talking.」

6、Omitted:

「Must the relationshipbetween the mechanism and the phenomenon be regular? This is an area of activediscussion (DesAutels 2011; Andersen 2011, 2014a,b; Krickel 2014). MDCstipulate that mechanisms are regular in that they work 「always or for the mostpart in the same ways under the same conditions」 (2000: 3). Some haveunderstood this (incorrectly in our view) as asserting that there are nomechanisms that work only once, or that a mechanism must work significantly morethan once in order to count as a mechanism.

「Some argue thatmechanisms have to be regular in this factual sense (Andersen 2014a,b); i.e.,repeated on many occasions (see Leuridan 2010). This view would seem to requirea somewhat arbitrary cut-off point in degree of regularity between things thattruly count as mechanisms and those that do not. Some mechanists (Bogen 2005;Glennan 2009) argue that there is no difficulty applying the term 「mechanism」to one-off causal sequences, as when an historian speaks of the mechanism thatgave rise to World War I. Other mechanists argue that the type-tokendistinction is too crude a dichotomy to capture the many levels of abstractionat which mechanism types and tokens might be characterized (Darden 1991).

「It is possible,however, to read the MDC statement as asserting, not a factual kind ofregularity, but as a counterfactual kind of near-determinism: were all theconditions the same, then the mechanism would likely produce the samephenomenon, where 「likely」 accommodates mechanisms with stochastic elements.

「While the MDC accountleaves open the possibility that some mechanisms are stochastic, it clearlyrules out mechanisms that usually fail to produce their phenomena. Skipper andMillstein (2005) press this point to argue that the MDC account cannotaccommodate the idea that natural selection is a mechanism. If, as Gould (1990)argued, one could not reproduce the history of life by rewinding the tapes andletting things play forward again, then natural selection would not be an MDCmechanism (see also Section 2.6 below). It is unclear why MDC would allow forthe possibility of stochastic mechanisms and rule out, by definition, thepossibility that they might fail more often than they work. Whether anybiological mechanisms are truly irregular in this sense (i.e., all the causallyrelevant factors are the same but the product of the mechanism differs) is aseparate question from whether they are mechanisms simpliciter (see Bogen 2005;Machamer 2004; Steel 2008 develops a stochastic account of mechanisms).

6、 Omitted:

「While the MDC accountleaves open the possibility that some mechanisms are stochastic, it clearlyrules out mechanisms that usually fail to produce their phenomena. Skipper andMillstein (2005) press this point to argue that the MDC account cannotaccommodate the idea that natural selection is a mechanism. If, as Gould (1990)argued, one could not reproduce the history of life by rewinding the tapes andletting things play forward again, then natural selection would not be an MDCmechanism (see also Section 2.6 below). It is unclear why MDC would allow forthe possibility of stochastic mechanisms and rule out, by definition, thepossibility that they might fail more often than they work. Whether any biologicalmechanisms are truly irregular in this sense (i.e., all the causally relevantfactors are the same but the product of the mechanism differs) is a separatequestion from whether they are mechanisms simpliciter (see Bogen 2005; Machamer2004; Steel 2008 develops a stochastic account of mechanisms).

「Krickel (2014)reviews the many different ways of unpacking the relevant notion of regularity(see also Andersen 2012). Her favored solution, 「reverse regularity,」 holdsthat there must be a generalization to the effect that, typically, when thephenomenon occurs, the mechanism was acting.」

7、Omitted:

「According to transmission accounts, causation involves thetransmission and propagation of marks or conserved quantities (Salmon 1984,1994; Dowe 1992). The most influential form of this view holds that two causalprocesses causally interact when they intersect in space-time and exchange someamount of a conserved quantity, such as mass. On this view, causation is local(the processes must intersect) and singular (it is fully instantiated inparticular causal processes), though the account relies upon laws ofconservation (Hitchcock 1995). Although this view inspired many of the newmechanists, and although it shares their commitment to looking toward sciencefor an account of causation, it has generally been rejected by new mechanists(though see Millstein 2006; Roe 2014).

「This view has been unpopular in part because it has littledirect application in nonfundamental sciences, such as biology. The causal claimsbiologists make usually don't involve explicit reference to conservedquantities (even if they presuppose such notions fundamentally) (Glennan 2002;Craver 2007). Furthermore, biological mechanisms often involve causation byomission, prevention, and double prevention (that is, when a mechanism works byremoving a cause, preventing a cause, or inhibiting an inhibitor) (Schaffer2000, 2004). Such forms of causal disconnection are ubiquitous in the specialsciences.」

8、Omitted:

「Glennan (1996, 2009) sees causation (at leastnon-fundamental causation) as derivative from the concept of mechanism: causalclaims are claims about the existence of a mechanism. The truth-maker for acausal claim at one level of organization is a mechanism at a lower level. In short,mechanisms are the hidden connexion Hume sought between cause and effect. Likethe Salmon-Dowe account, Glennan’s view is singular: particular mechanisms linkparticular causes and particular effects (Glennan forthcoming)

「This view has been charged with circularity: the concept ofmechanism ineliminably contains a causal element. However, Glennan replies thatmany accounts of causation (such as Woodward's 2003 account, see Section 2.3.4below) share this flaw. Furthermore, he argues that for at least allnon-fundamental causes, a mechanisms clearly explains how a given causeproduces its effect.

「Whether the analysis succeeds depends on how one deals withthe resulting regress (Craver 2007). As Glennan (2009) notes, the decompositionof causes into mechanisms might continue infinitely, in which case there is nopoint arguing about which notion is more fundamental, or the decompositionmight ground out in some basic, lowest-level causal notion that is primitiveand so not analyzable into other causal mechanisms. The latter option mustconfront the widely touted absence of causation in the theories of fundamentalphysics (Russell 1913); at very small size scales, classical conceptions ofobjects and properties no longer seem to apply, making it difficult to see whatcontent is left to the idea that there are mechanisms at work (see also Teller2010; Kuhlman and Glennan 2014).」

9、Omitted:

「Still other mechanists, such as Bogen (2005, 2008a) andMachamer (Machamer 2004), embrace an Anscombian, non-reductive view thatcausation should be understood in terms of productive activities (see also theentry on G.E.M. Anscombe). Activities are kinds of causing, such as magneticattraction and repulsion or hydrogen bonding. Defenders of activity-basedaccounts eschew the need to define the concept, relying on science to say whatactivities are and what features they might have. This view is a kind of causalminimalism (Godfrey-Smith 2010). Whether an activity occurs is not a matter ofhow frequently it occurs or whether it would occur always or for the most partin the same conditions (Bogen 2005).

「This account has been criticized as vacuous because itfails to say what activities are (Psillos 2004), to account for therelationship of causal and explanatory relevance (Woodward 2002), and to markan adequate distinction between activities and correlations (Psillos 2004),though see Bogen (2005, 2008a) for a response. Glennan (forthcoming) arguesthat these problems can be addressed by recognizing that activities in a mechanismat one level depend on lower-level mechanisms. (See also Persson 2010 for acriticism of activities based on their inability to handle cases of polygeniceffects.)」

10、Omitted:

」Lastly, some new mechanists, particularly those interestedin providing an account of scientific explanation, have gravitated toward acounterfactual view of causal relevance, and in particular, to themanipulationist view expressed in Woodward (2001, 2003) (see, e.g., Glennan2002; Craver 2007). The central commitment of this view is that models ofmechanisms describe variables that make a difference to the values of othervariables in the model and to the phenomenon. Difference-making in thismanipulationist sense is understood as a relationship between variables inwhich interventions on cause variables can be used to change the value ofeffect variables (see the entry on causation and manipulability).

「Unlike the views discussed above, this way of thinkingabout causation provides a ready analysis of explanatory relevance thatcomports well with the methods for testing causal claims. Roughly, one variableis causally relevant to a second when there exists an ideal intervention on thefirst that changes the value of the second via the change induced on the first.The view readily accommodates omissions, preventions, and doublepreventions—situations that have traditionally proven troublesome forproduction-type accounts of causation. In short, the claim that C causes Erequires only that ideal interventions on C can be used to change the value ofE, not that C and E are physically connected to one another. Finally, this viewprovides some tools for accommodating higher-level causal relations and thenon-accidental laws of biology. On the other hand, the counterfactual accountis non-reductive (like the mechanistic view), and it inherits challenges facedby other counterfactual views, such as pre-emption and over-determination whichare common in biological mechanisms (see the entry on counterfactual theoriesof causation).」

11、Omitted:

「Wimsatt (1997) contrasts mechanistic organization withaggregation, a distinction that mechanists have used to articulate how theparts of a mechanism are organized together to form a whole (see Craver 2001b).Aggregate properties are properties of wholes that are simple sums of theproperties of their parts. In aggregates, the parts can be rearranged andintersubstituted for one another without changing the property or behavior ofthe whole, the whole can be taken apart and put back together withoutdisrupting the property or behavior of the whole, and the property of the wholechanges only linearly with the addition and removal of parts. These features ofaggregates hold because organization is irrelevant to the property of thewhole. Wimsatt thus conceives organization as non-aggregativity. He alsodescribes it as a mechanistic form of emergence (see Section 4.2 below).

「Mechanistic emergence is ubiquitous—truly aggregativeproperties are rare. Thus mechanists have tended to recognize a spectrum oforganization, with aggregates at one end and highly organized mechanisms on theother. Indeed, many mechanisms studied by biologists involve parts and causingsall across this spectrum. (For further discussion of mechanistic emergence inrelationship to other varieties, see Richardson and Stephan 2007.)」

12、Omitted:

「Following Wimsatt, mechanists have detailed several kindsof organization characteristic of mechanisms. A canonical list includes bothspatial and temporal organization. Spatial organization includes location,size, shape, position, and orientation; temporal organization includes theorder, rate, and duration of the component activities. More recently,mechanists have emphasized organizational patterns in mechanisms as a whole. Bechtel,for example, discusses how mathematical models, and dynamical models inparticular, are used to reveal complex temporal organization in interactivemechanisms (Bechtel 2006, 2011, 2013b). Some argue that dynamical models pushbeyond the limits of the mechanistic framework (e.g., Chemero and Silbestein2008 and, at times, Bechtel himself; see Kaplan and Bechtel 2011). Others arguethat dynamical models are, in fact, often merely descriptive (i.e.,non-explanatory models) or, alternatively, that they are used to describe thetemporal organization of mechanisms (Kaplan and Bechtel 2011; Kaplan 2012).

「Mechanists have also recently borrowed from Alon’s (2006;Milo et al. 2002) work on network motifs, repeated patterns in causal networks,to expand the vocabulary for thinking about abstract patterns of organization(Levy 2014; Levy and Bechtel 2012). Understanding how parts compose wholes islikely to be a growth area in the future of the mechanistic framework. (Forsome other recent additions, see Kuorikoski and Ylikoski 2013; Kuhlmann 2011;Glennan forthcoming.)」

13、Omitted:

「Woodward’s (2001, 2002, 2011, 2014) counterfactualdefinition of a mechanism (which is indirectly specified via an account ofmechanistic models), as well as a descendant elaborated by Menzies (2012),require that models of mechanisms be modular. This means, roughly, that itshould be physically possible to intervene on a putative cause variable in amechanism without disrupting the functional relationships among the othervariables in the mechanism. In terms of structural equation models inparticular, this means that one should be able to replace the right-hand sideof an equation in the model with a particular value (i.e., set the left-handvariable to a value) without needing to change any of the other equations inthe model. This is intended to formally capture the sense in which mechanism「Steel (2008) appeals to a somewhat weaker form of modularity in hisprobabilistic analysis of mechanisms—one that follows directly from Simon's (1996[1962]) idea of nearly decomposable systems. On Simon's view, the parts of amechanism have more and stronger causal relations with other components in themechanism than they do with items outside the mechanism. This gives mechanisms(and parts of mechanisms) a kind of 「independence」 or 「objecthood」 definedultimately in terms of the intensity of interaction among components. Grush(2003), following Haugeland (1998), develops an idea of modularity in terms ofthe bandwidth of interaction, where modules are high-bandwidth in theirinternal interactions and low-bandwidth in their external interactions. On thisview, modularity is not an all-or-none proposition but a matter of degree;mechanisms are only nearly decomposable. Craver (2007) argues that such ageneric notion fails to account for the relevance of different causalinteractions for different mechanistic decompositions; what counts as a part ofa mechanism can only be defined relative to some prior decision about what onetakes the mechanism to be doing. For criticisms of modularity, see Mitchell(2005) and Cartwright (2001, 2002).」

14、Omitted:

「Fagan (2012, 2013) emphasizes the interdependentrelationship between parts of a mechanism. Components in a mechanism, shepoints out, often form a more complex unit by virtue of the individualproperties that unite them—their 「meshing properties」; the complex unit thenfigures into the mechanism's behavior. This interdependentrelationship—jointness—is exemplified by the lock-and-key model of enzyme action.Fagan applies this notion to research on stem cells (Fagan 2013) but arguesthat it is a general feature of experimental biology (Fagan 2012).」

15、Omitted:

「Many mechanists emphasize the hierarchical organization ofmechanisms and the multilevel structure of theories in the special sciences(see especially Craver 2007, Ch. 5). Antecedents of the new mechanism focusedalmost exclusively on etiological, causal relations. However, the new emphasison mechanisms in biology and the special sciences demanded an analysis ofmechanistic relations across levels of organization.

「From a mechanistic perspective, levels are not monolithicdivides in the furniture of the universe (as represented by Oppenheim andPutnam 1958), nor are they fundamentally a matter of size or the exclusivity ofcausal interactions within a level (Wimsatt 1976). Rather, levels of mechanismsare defined locally within a multilevel mechanism: one item is at a lower levelof mechanisms than another when the first item is a part of the second and whenthe first item is organized (spatially, temporally, and actively) with theother components such that together they realize the second item. Thus, themechanism of spatial memory has multiple levels, some of which include organssuch as the hippocampus generating a spatial map, some of which involve thecellular interactions that underlie map generation, and some of which involvethe molecular mechanisms that underlie those cellular interactions (Craver2007). For more on levels, see Section 4.2 below.」

16、Wikipedia:

「Concatenation In formallanguage theory and computer programming,stringconcatenation is the operation of joining character strings end-to-end. Forexample, the concatenation of 「snow」 and 「ball」 is 「snowball」. In some but notall formalizations of concatenation theory, also called string theory, stringconcatenation is a primitive notion.」

17、The Parts, Activities,Organization in Mechanism in Figure 1, that is,

can be represented asfollows:

x1ψ-ing→x2ψ-ing→x4ψ-ing + x1ψ-ing→x2ψ-ing→x3ψ-ing→x4ψ-ing+ x1ψ-ing→x3ψ-ing→x2ψ-ing→x4ψ-ing)+ x1ψ-ing→ x3ψ-ing→x4ψ-ing, each of which is apathway for Causings to pass through, ignoring the feedback loop between x2ψ-ingand x3ψ-ing for the sake of simplicity.

Simplifying, we obtain:

(x1ψ→x2ψ→x4ψ + x1ψ→x2ψ→x3ψ→x4ψ+ x1ψ→x3ψ→x2ψ→x4ψ + x1ψ→x3ψ→x4ψ)-ing

Adding a null part x0, weobtain:

(x1ψ→x2ψ→x0ψ→x4ψ + x1ψ→x2ψ→x3ψ→x4ψ+ x1ψ→x3ψ→x2ψ→x4ψ + x1ψ→x3ψ→x0ψ→x4ψ)-ing

Grouping, aggregating, concatenating,separating and replacing notations where necessary, we obtain:

([x1ψ→x2ψ]→x0ψ→x4ψ + [x1ψ→x2ψ]→x3ψ→x4ψ + [x1ψ→x3ψ]→x2ψ→x4ψ + [x1ψ→ x3ψ]→x0ψ→x4ψ)-ing

= ([x1ψ→x2ψ]→[x0ψ+x3ψ]→x4ψ + [x1ψ→x3ψ]→[x2ψ+x0ψ]→x4ψ)-ing

= ([[x1※ψ→x2※ψ]→[x0※ψ+x3※ψ] + [x1※ψ→x3※ψ]→[x2※ψ+x0※ψ]]→x4※ψ)-ing

= ([x1※ψ→x2※ψ→[x0+x3]※ψ + x1※ψ→x3※ψ→[x2+x0]※ψ→x4※ψ])-ing

= (x1(x2([x0+x3]※ψ)ψ)ψ + x1(x3([x2+x0]※ψ)ψ)ψ→x4※ψ)-ing

= (x1([x2([x0+x3] + x3([x2+x0]]※ψ)ψ)ψ→x4※ψ)-ing

= (x1([x2([x0+x3] + x3([x2+x0]](x4※ψ)ψ)ψ)ψ)-ing

where [ ] and + shouldread respectively 「grouping」 and 「aggregation」.

As a result,S = (x1([x2([x0+x3] + x3([x2+x0]](x4 in S※Ψ-ing, and Ψ-ing = ψ)ψ)ψ)ψ)-ing.It is noticed that when there is a single beginning and a single ending S canbe represented as a single formula. Even if there are more than one beginningor one ending we can create an artificial one like a null part, that is, a nullsingle beginning or ending.

Clearly, when ψ is replaced by ψi, i = 1,2, ... the representation should be modified accordingly.

18.◇H-D,that is, a single incident, is a mechanism with only one xψ-ing, where x =definitions of the situation and ψ-ing =structure-cum-society. The explanandum x (as data) is obviously the phenomenon itself, and is embedded in themechanism xψ-ing.

◇H-D―CE, that is, a culturally explained incident, is amechanism with x2(x1ψ)ψ-ing, where x1 = definitions of the situation, x2= grounds for interaction and ψ-ing =structure-cum-society. The explanandum x1 (as data) is in the phenomenon.

A sequence ofculturally explained incidents, that is, ◇H-D1―CE1, ◇H-D2―CE2, ..., can be proved to be a mechanism in thestraightforward manner.

19.Hedstrom,Peter. 2005. Disserting the Social—On thePrinciples of Analytical Sociology. Cambridge University Press: Cambridge,New York, Melbourne, Madrid, Cape Town, Singapore, Sao Paulo

20. Popper, K.R. 1994. 「Models, Instruments, and Truth: The Status of the RationalityPrinciple in the Social Sciences,」 in TheMyth of the Framework: In Defence of Science and Rationality, ed. K. R.Popper. London: Routledge, pp. 154-184.

21. Wikipedia:

「A jigsaw puzzle is a tiling puzzle that requires the assembly of oftenoddly shaped interlocking and tessellating pieces. Each piece usually has asmall part of a picture on it; when complete, a jigsaw puzzle produces acomplete picture. [...]」

22. See commentsfrom proponents and opponents in orgtheory.net, a website for informalexchanges of opinions and ideas, for example, 「More words on critical realism:Getting clear on the basis,」 written by Omar, September 14, 2013 at 7:48 pm

23. Steinmetz,George. 1998. 「Critical Realism and Historical Sociology. A Review Article,」 inComparative Studies in Society andHistory, Vol. 40, Issue 1, pp. 177-186.

24. Bhaskar,Roy. 1997. A Realist Theory of Science.1997 edition by Verso, p. 13

25. 「[...] Iwill define a 『closed system』 simply as one in which a constant conjunction ofevents obtains; i.e. in which an event of type a is invariable accompanied byan event of type b.」 Bhaskar, op. cit.,p, 70.

26. Alfred NorthWhitehead’s Process and Reality issubtitled 「An essay in cosmology」

27. 「Depthrealism」 is the title given to Bhaskar’s original table in Steinmetz’s articlementioned earlier.

28. 「Ontology,it should be stressed, does not have as its subject matter a world apart fromthat investigated by science. Rather, its subject matter just is that world,considered from the point of view of what can be established about byphilosophical argument.」 Bhaskar, op.cit., p.70.

29. Steinmetz, op. cit., p.177

30. Bhaskar, op, cit., p.17.

31. See footnote25.

32. Steinmetz, op. cit., p. 176.

33. Bhaskar, op, cit., p.18。

34. Ditto, p.50. Steinmetz: 「Generative mechanisms are「tendencies」 rather than 「powers」 because they are not just potentialities butpotentialities that may be exercised without being manifested.」 Steinmetz, op. cit., p.177, footnote 19.

35. A Chineseproverb 「天道酬勤」(a literal translation is 「Heaven’s way is toreward the industrious (men)」) illustrates these two presuppositions. Heaven’s way（天道）is first of all a power（力量）, that is, the power to reward men. But Heaven doesnot reward itself, so it is realized but incompletely manifest if there is noman to reward. It is however always poised to do so, and so a tendency（傾向）.

36. Newton’sLaws as tendencies are out there as stock knowledge waiting to be possessed,exercised, realized or perceived by men. Yet, it does not mean they are easy topossess—very likely less than half of the living adults in the world can do so.They are therefore some sort of free-for-all but used-by-few stock knowledge,and we may call it 「God Knows」 (GK).

37. Bhaskar, op. cit., p.20

38. Elizabeth S. Clemens,「Organizational repertoires and institutional change: Women’s groups and thetransformation of U.S. politics, 1890-1920,」 in American Journal of Sociology, Vol. 98, No. 4 (Jan., 1993), pp.755-798.

39. Original footnote 22:

「A similarassumption is found in much neo-Marxist work on modern society that focuses onthe contradictions between institutions held to have different 「logics」 (e.g.,Alford and Friedland 1985; Block 1977; O'Connor 1973).」

本文為龍子湖公眾號原創，並已得到作者呂炳強老師授權。

讀者如有問題，歡迎通過以下郵箱討論：

intergaze@126.com

Mechanisms and Mechanismic Explanations

相關焦點

Clonal hematopoiesis: mechanisms driving dominance of stem ...

...Qi’s group reveals a new mechanism for sexual dimorphic...

2014年冷泉港亞洲會議:Mechanisms of Transmembrane Signaling

神經網絡中的注意機制(Attention Mechanism)

U.S.-China dialogue attracts global attention

...text: Enhanced Actions on Climate Change: China's Intended...

CRISPR helps identify gene that regulates coral's heat stress...

Full Text: China's Space Activities in 2016

GPRASP proteins are critical negative regulators of ...