https://orcid.org/0000-0003-3019-8348
![Sample our Humanities journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5939/TOC-Subject-Sample-2021-Humanities.jpg"})
ABSTRACT
James Woodward’s manipulationist counterfactual theory of explanation offers strong tools for an adequate approach to explanation endeavours. One of these tools is the notion of intervention, which serves as a guiding principle for identifying explanations as causal, thus preserving the unidirectionality of explanatory praxis. Nevertheless, in this paper, I argue that in some cases of explanation, this notion has a rather redundant role since it is either impossible to define or it can be replaced by other types of manipulations or both. I shall show these shortcomings of intervention in two particularly detailed analyses of an economic example pertaining to Okun’s law and the pendulum explanation. In analysing the former example, I contend that given the complex causal situation of economic phenomena, the notion of intervention is inept at helping explain them. In the latter example, I show that the notion of intervention is readily replaceable by other types of variation—conceptual or mathematical. Finally, I shall end this article by examining the case of pendulum explanation in light of Woodward’s new approach to explanation direction. In the end, the conclusion is that the variables of the pendulum law are in a symmetrical relationship.
Acknowledgements
I am grateful to the editors and three anonymous reviewers who shared with me useful suggestions and constructive criticism. I am also thankful to my doctoral advisor doc. Lukáš Bielik who provided me with valuable advice and knowledge.
Disclosure Statement
No potential conflict of interest was reported by the author(s).
Notes
1 This claim is not universally accepted. See Strevens (Citation2018, 109–111), for instance. Even though I find Strevens’ analysis very illuminating and stimulating, I cannot agree that even high-abstract relations, such as the one in Euler’s explanation, can be traced to some fundamental causal influences.
2 Okun’s law in its full description: ‘Okun’s Law is a short-run relationship between the deviation of output from potential and the deviation of unemployment from its natural rate … shifts in aggregate demand cause movements in output, which, in turn, drive fluctuations in the labour market: firms hire and fire workers to accommodate output changes, and these actions affect employment and unemployment.’ (Ball, Leigh, and Loungani Citation2017, 1415).
3 This is why some philosophers are against expunging the notion of causality from the explanation process. (See Salmon Citation1989, 24).
4 I am abstaining from answering the question of whether there is a possible physical intervention that could be applied on either variable. I take it that if a conceptually well-defined (or ideal) intervention is impossible to be devised, then a physical intervention is impossible to be devised a fortiori. I will regard these variables only within the boundaries of an ideal model.
5 What I wanted to show in this section is that Okun's law can be derived from two underlying equations by the operation of substitution. For employment, we have this simplified relationship E = γY (1), where E is employment, Y is output (In my manuscript I have chosen the letter O to represent the output) and γ is the inverse of the elasticity of output with respect to labour. For unemployment, we have this simplified equation: U = δE (2), where U is unemployment and δ represents shifts in the labour force. From these two equations, we can derive by substitution Okun’s law: U = βY (3), where β depends on the values from the two previous parameters. Since we can derive Okun's law by substituting the first equation (employment) in the second one I surmise that there obtains a transitive relationship between U and Y (in my manuscript O).
6 One of the reviewers pointed out that the Woodwardian counterfactual should have the form: I →O → E → U. My understanding of the relationship I → X → Y is that each letter represents a variable. Hence, I allowed myself to substitute concrete events for each variable so that the relationship is as follows: O → E → U.
Also, I think that Woodward's account permits even an event as output O to figure as an intervention since the intervention itself does not have to make a reference to notions like human agency – it can be understood in a nonanthropomorphic way (Woodward Citation2003, 14). But I also agree with the reviewer that the second form is a potential form of a Woodwardian manipulationist counterfactual. However, in this case, the intervention I could represent the shifts in aggregate demand which in turn would affect output O. Nevertheless, I think this second form has the same issue as the first one since in the second form we cannot definitively conclude that the only pathway through which I affects U is O → E → U.
7 Woodward (Citation2018a, 123) shows a somewhat similar example where he tries to argue that if there is no clear notion of an intervention, like in the case of a causal change of the dimensionality of space, then this change which explains the (non)stability of orbits can be potentially replaced by a mathematical one.
8 To decide on the exact form/course of the mathematical variation and the explanation direction of variables (U) and (O) is beyond the scope of this paper. However, for a succinct elaboration on how counterfactual non-causal explanations might be reconstructed in biology and potentially in economics as optimality models/ explanations, see Rice (Citation2015, 19–21).
9 As a disclaimer, I want to add that my goal of this section was not to show that if we are unable to devise a well-defined intervention, then the causal relations within the science field of economics cannot be observed through interventions altogether. My whole analysis in this section should only apply to the case of Okun’s law.
10 By challenge, I mean how to decide on the issue of whether this explanation is a causal one or not.
11 A relation or, in this case, a law that describes the system only from an atemporal point of view. Opposed to this is the diachronic relation that can describe the state of a system in an earlier or later time.
12 See Woodward’s first chapter in (Citation2003).
13 Explanation asymmetry between factors – which represents a fixed explanation priority for each factor – is, for many authors, a necessary condition for explaining. (See Lewis Citation1987, Salmon Citation1989, Lange Citation2019a, Woodward Citation2018a)
14 Consider this approach. Let us have: K│¬E → N, which reads as given that K, the non-Eulerian configuration of bridges ¬E explains why it is impossible to walk over all bridges in one go N. An alternative notation: K│N → ¬E, is also possible, but I do not think that the latter represents an explanation of the fact ¬E by the fact N. This is because when we ask why the bridge arrangement is non-Eulerian, we do not need to refer to N as long as K is given. It is sufficient to refer to K since it can explain that ¬E without N.
15 I understand the relationship between the variables in the pendulum law as a synchronic mathematical one.
16 General abnormic laws are equivalent to a conjunction of special abnormic laws (Bromberger Citation1966, 99). For a more detailed and succinct account of Bromberger’s take on the asymmetry issue of the DN model, see Salmon (Citation1989, 43–46)
17 My point was that anything resembling a pendulum would have the same length, whether it would be oscillating (moving) or not. Even something that is not a pendulum, i.e. a suspended wire with a bob lacking a non-rigid pivot point. Without a nonzero value, the period of a pendulum at (T = 0) is not a valid descriptive state of this system since, as a single value, it is confounding about the real state of the system – that is, whether or not it can be described as a real pendulum and not just a rigid suspended bob. The static equilibrium of a pendulum is not describable with just the variable (T) because one would need to specify the forces or momenta exerted on the bob that are, in fact, responsible for this state. Therefore, I think the result L(0) = 0 is a hint pointing out the confounding nature of setting the variable (T) to zero in the pendulum law.
As a final remark, I would like to point out the incoherency in the abnormic pendulum law: ‘No pendulum has a suspension of length L unless it oscillates with period T.’ and Salmon’s conclusion about it. According to Salmon (Citation1989, 45), this law is just not true because ‘a pendulum even in resting position (T = 0) would have such a length’. But this conclusion is not valid, for it is not that a pendulum, even in resting position (T = 0), would have such a length it could have any length. If Salmon refers to a certain length of a pendulum, then the value (T = 0) is not a unique identifier of such a concrete length. A pendulum could be of any length whilst the variable (T) equals zero.
18 This can be seen in his paper (Citation1966), where he emphasises logic and its interplay with natural language in elucidating counterfactual statements.
19 Although his ideas are loosely based upon the work of Eugen Wigner (Citation1967). As he states in his papers. See Woodward (Citation2018b, Citation2020).
20 Woodward writes (Citation2018b, 165): ‘If a regularity or non-independence is present among the initial conditions, one would expect that there is some further explanation of this, which traces the dependency or regularity back to earlier conditions satisfying the randomness requirement.’ Nevertheless, I do not think that this idea applies to the case of the pendulum explanation because the (simple) pendulum as a system is a not-evolving one – it only concerns changes in variables (L) and (T) at one and the same time.
21 I think this criterium could be a relevant addition to the few others that try to elucidate the nature of the non-causal explanation. Notwithstanding the possibility of non-causal explanations that (could) have a fixed explanation priority of facts, I suppose that a sufficient condition for denoting an explanation as non-causal is to show that the variables or factors constituting it are in a symmetrical relationship. One can achieve this by showing that the factors are not asymmetrical – there is no cause-effect distinction between factors – or one could show that there obtains no time asymmetry between them. One such approach that uses the so-called Russellian criteria is offered by Reutlinger (Citation2018, 89). There are four of these criteria from which, if only one is not met, the explanation must be taken as non-causal. The other account that tries to make sense of the explanation direction while leaving open the possibility of symmetry explanations is Hempel’s (Citation1965, 352). He distinguishes between two types of laws: laws of succession and laws of coexistence. These should reflect the potential synchronic and diachronic relations that obtain between various variables (representing facts), enabling us to determine the (bi)directionality of the explanation.
22 This conclusion may not sit very well with those philosophers of science that maintain the causal asymmetry requirement for explanatory endeavours. Still, I am more inclined to adhere to Russell’s conclusion that causation in and of itself does not belong to the modern scientific vocabulary – ‘that it is a relic of a bygone age’ (Russell Citation1913, 1). For a more current treatment of Russell’s arguments, specifically the Directional argument, see Farr and Reutlinger (Citation2013).