Skip to main content
Log in

Mutual manipulability and causal inbetweenness

  • Published:
Synthese Aims and scope Submit manuscript

Abstract

Carl Craver’s mutual manipulability criterion aims to pick out all and only those components of a mechanism that are constitutively relevant with respect to a given phenomenon. In devising his criterion, Craver has made heavy use of the notion of an ideal intervention, which is a tool for illuminating causal concepts in causal models. The problem is that typical mechanistic models contain non-causal relations in addition to causal ones, which is why the standard concept of an ideal intervention is not appropriate in that context. In this paper, I first show how top-down interventions in mechanistic models violate the conditions for ideal interventions. Drawing from recent developments in the causal exclusion literature, I then argue for extended interventionism better suited for the purposes of the new mechanist. Finally, I show why adopting such an extended account leads to the surprising consequence that an important subset of mechanistic interlevel relations comes out as causal.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3

Similar content being viewed by others

Notes

  1. As an anonymous reviewer points out, there is bound to be some interest-relativity as to how detailed one’s mechanistic story of a phenomenon should be. The situation is analogous to causal modeling, where the appropriate grain of one’s variables depends on the explanatory task at hand.

  2. Throughout this paper I will engage in variable-talk. Variables are very flexible in that they can be thought to correspond to all kinds of things. For example, if one wishes, it is possible to think of variables as corresponding to events, with the following two values: {occurs, doesn’t occur} For many, this would be the appropriate interpretation for the purposes of causal analysis. However, the values of variables needn’t correspond to events, or be binary, which is welcome given that there are many situations in which it is useful to have e.g. many-valued variables. Later I will also talk as if the values of variables would engage in supervenience or realization relations, in which case the variables are best interpreted as corresponding to properties.

  3. In the mechanistic context, Soom (2012) has argued that the relation between S’s \(\psi \)-ing and X\(_{1,\ldots ,n}\)’s \(\phi _{1,\ldots n}\)-ing should be understood in terms of ‘strong supervenience’, while Harbecke (2013) has put forward a mechanistic modification of what is known as ‘coordinated multiple-domain supervenience’. Both types of supervenience were originally articulated by Kim (cf. 1984; 1988).

  4. Note that the overall behaviour of a mechanism is usually characterized in extrinsic terms, as a kind of input-output regularity. The advocates of mechanistic supervenience, then, argue for a very similar view as those who hold that dispositions supervene on their ‘bases’. This is worth mentioning because the supervenience thesis about dispositions has been challenged (cf. Mumford 1994, McKitrick 2003).

  5. Here I am following Woodward (2014, p. 29–31) in making the ontologically and empirically plausible assumption that an intervention on a supervening variable \(\psi \) is a direct and simultaneous intervention on its supervenience base SB(\(\psi \)): there is just one intervention that is changing both. For those who do not wish to grant this assumption, it is worth noting that the relationship between \(\psi \) and SB(\(\psi \)) also violates the interventionist requirement that one should be able to vary the value of each variable in one’s model via an intervention while holding the values of any of the other variables in the model fixed at any of the values within their normal range via independent interventions (Hausman and Woodward 1999; Woodward 2014, see also footnotes 9 and 10). If needed, my conclusion about the lack of ideal interventions on \(\psi \) w.r.t SB(\(\psi )\) could be reached trough that route too. In what follows, I will continue to assume that an intervention on \(\psi \) is also an intervention on SB(\(\psi \)). I thank anonymous reviewer for drawing my attention to this issue.

  6. Indeed, as David Papineau has pointed out (personal communication), it is even possible to understand realization as the converse of supervenience: \(\phi \) realizes \(\psi \) if and only if \(\psi \) supervenes on \(\phi \). Under this analysis, it is hardly surprising if both relations turn out to be equally problematic for ideal interlevel interventions. But even under the more complex definitions of realization put forward in recent debates, which often make no overt reference to supervenience, it is still plausibly the case that changes in \(\psi \) require changes in \(\phi \). Note that I am here intentionally working with a very abstract characterization of the realization relation. The reason is that there is currently considerable dispute as to the ‘appropriate’ type of realization relation in the mechanistic context (Polger 2010). In this dispute, Craver himself seems to side with those advocating the ‘dimensioned’ concept against the ‘flat’ one (Craver 2007, p. 212). But it is also questionable whether there is a genuine disagreement between these two notions to begin with (Endicott 2011).

  7. The argument in the previous section also shows why the worries raised by Leuridan (2012, p. 407–409) about interlevel interventions satisfying conditions (I1)–(I4) are premature. There are no such interlevel interventions because conditions (I1) or (I2) will be violated as the result of the close-knitted metaphysical relationship between the mechanism as a whole and its components-taken-together, such as supervenience or realization. Leuridan’s argument is that if we assume that Craver’s interlevel interventions satisfy conditions (I1)–(I4), then it very much looks as if those interventions pick out causal relevance relations. My answer here is that we should not make such an unrealistic assumption in the first place. In any case, Leuridan then goes on to ask whether Craver could use a parthood criterion to argue that mechanistic interlevel interventions are not causal. In order to do that, he thinks, Craver would need to be able to assume that (i) ‘if X is part of S, then an intervention I on X directly changes S’ whilst denying that (ii) ‘if X is part of S, then an intervention I on X’s \(\phi \)-ing directly changes S’s \(\psi \)-ing’. The problem is, according to Leuridan, that it is hard to make the former assumption without making the latter. As it happens, in Sect. 6.1 I will demonstrate why what Leuridan claims is difficult for Craver is not difficult at all: the key move is to understand mutual manipulability as involving three variables. For further discussion of Leuridan’s argument, see Footnotes 16 and 18.

  8. Note that the problem does not depend on whether the mutual manipulability criterion is interpreted ‘metaphysically’ or ‘epistemologically’ (cf. Schindler 2013). Either way, the interlevel interventions turn out non-ideal. Moreover, the following question would remain even under the weaker epistemological reading: what is the metaphysical structure that grounds top-down and bottom-up experiments?

  9. One further assumption here and elsewhere in this paper is that a supervenient variable cannot be causally related with its supervenience base. This is trivially the case if the relationship between the two violates condition (I1) for ideal interventions, as I think it does. But again the same conclusion can be achieved by observing that the two variables violate the independent fixability condition (see footnotes 5 and 10).

  10. The debate gets more complex than this. For instance, it is usually framed in terms of the independent fixability condition that many (cf. Baumgartner 2010; Halpern and Hitchcock 2011; Woodward 2014) take to be an important interventionist assumption and that is also discussed in footnotes 5 and 9. A roughly similar assumption can be found in the work of Craver (2007, p. 156–157), although it is not clear whether Craver’s version of the condition is equivalent to those put forward by others (cf. Woodward 2014, p. 14). (I thank anonymous referee for pointing out the apparent lack of equivalence.) For the purposes of this paper, however, we can happily continue to frame the debate in terms of the conditions for ideal interventions.

  11. This is just how Lewis treats counterfactuals with impossible antecedents. The motivation that Lewis gives for this involve the intuition, which some might have, that if something impossible were to be true, then anything could be true (Lewis 1973, p. 24). Another thing worth noting is that Woodward himself requires that, in order for X to cause Y, there must be an ideal intervention on X with respect to Y (Woodward and Hitchcock 2003; Woodward 2005). But it is not clear whether this commitment is essential for the interventionist programme.

  12. One might worry that the proposed extension is ad hoc, and Woodward himself (2014, p. 30) concedes that his conditions for extended interventions are ‘cumbersome’. But I believe the only reason (EI1) and (EI2) sound ad hoc is that they weren’t included in the original set of complex conditions used to formulate interventionism in the first place. The crucial test here is whether they capture the scientific practice, which I’ve argued they do. I thank an anonymous reviewer for drawing my attention to this worry.

  13. Baumgartner (2013) argues that scientists might want to regard D and SB(D) as competitors if they suspect that D does not reduce to SB(D). If there would be some sort of primitive and inexplicable supervenience relation between the two, then we might actually see scientists requiring interventions on D with respect to R that would at the same time leave SB(D) unchanged. Perhaps this is so. But the reason is the suspect nature of the supervenience relation between D and SB(D) and not because this is the correct way to do causal modeling.

  14. I thank Carl Craver for suggesting this name for my approach, as well as for the many valuable comments he provided regarding the ideas put forward in this section.

  15. I am not excluding the possibility that a researcher might sometimes wiggle the value of some \(\phi _{i}\) and detect changes in the value of the input variable \(\psi _{in}\). This type of case might occur, for example, if the \(\phi _{i}\) in question is among the variables on which \(\psi _{in}\) supervenes. However, even in this case it is plausible that the researcher would also require changes in \(\psi _{out}\) under the intervention on \(\phi _{i}\).

  16. This three-variable nature of mutual manipulability is missed by Leuridan (2012), whose argument was discussed earlier in Footnote 7. As we recall, his claim was that, in order to argue that mechanistic interlevel relations are not causal, Craver would have to maintain that (i) an intervention on part X of S directly changes S while simultaneously denying that (ii) an intervention on X’s \(\phi \)-ing directly changes S’s \(\psi \)-ing—something Leuridan thought Craver would have a hard time doing. However, the unpacking of S’s \(\psi \)-ing as involving two variables and the consequent interpretation of mutual manipulability as a three-variable affair show why one could easily hold (i) while denying (ii). The issue is a red herring. Why mechanistic interlevel relations come out as causal isn’t to do with the difficulty of holding the above combination of beliefs; it’s to do with what is a plausible account of interlevel interventions. My argument is that it is the extended one. See also Footnote 18.

  17. Or violates other important interventionist assumptions. See footnotes 5, 9 and 10.

  18. An anonymous reviewer asks what the relationship between constitution and causality is in the account that I’m giving. Elsewhere (unpublished) I have developed an approach to constitution under which parts must be causally in between the inputs and outputs defining the phenomenon for which the whole of which they are parts is responsible. My view is that this is a natural step to take if one is already willing to decide issues of constitutive relevance in terms of mutual manipulability. An obvious implication of this is that I do not accept the view that constitutive relations cannot be causal. On the contrary, I have outlined a perfectly good way in which there is a causal relationship between a mechanism’s behaviour as a whole and the individual behaviours of its components, even though the latter are constituents in the former. As the consequence of this, putative counterexamples to views that Craver may hold, such as the case of endosymbiosis discussed by Leuridan (2012, p. 412), are not counterexamples at all from my point of view.

  19. In practical terms, investigating the causal contribution of some single voltage-dependent gate is clearly unrealistic. A real-life researcher would be likely to intervene on a number of voltage-dependent gates in a given region and see whether that results in changes in the output (real-life interventions are most of the time ham-fisted). But recall the counterfactual element in interventionism. The appropriate question here is: would the neuron release a signal if an ideal intervention were to change the status of the voltage-dependent gate under investigation, whilst the values of various off-path variables, including those corresponding to the statuses of the other voltage-dependent gates in the relevant region, would be held fixed.

References

  • Baumgartner, M. (2010). Interventionism and epiphenomenalism. Canadian Journal of Philosophy, 40(3), 359–383.

    Article  Google Scholar 

  • Baumgartner, M. (2013). Rendering interventionism and non-reductive physicalism compatible. Dialectica, 67(1), 1–27.

    Article  Google Scholar 

  • Bechtel, W., & Abrahamsen, A. (2005). Explanation: A mechanist alternative. Studies in History and Philosophy of Science Part C, 36(2), 421–441.

    Article  Google Scholar 

  • Bechtel, W., & Richardson, R. C. (1993). Discovering complexity: Decomposition and localization as scientific research strategies. Princeton: Princeton University Press.

    Google Scholar 

  • Craver, C. F. (2007). Explaining the brain. Oxford: Oxford University Press.

    Book  Google Scholar 

  • Craver, C. F., & Bechtel, W. (2007). Top-down causation without top-down causes. Biology and Philosophy, 22(4), 547–563.

    Article  Google Scholar 

  • Cummins, R. C. (2000). “How does it work” versus “what are the laws?”: Two conceptions of psychological explanation. In F. Keil & R. A. Wilson (Eds.), Explanation and cognition (pp. 117–145). Cambridge: MIT Press.

  • Endicott, R. P. (2011). Flat versus dimensioned. Journal of Philosophical Research, 36, 191–208.

    Article  Google Scholar 

  • Gillett, C. (2013). Constitution, and multiple constitution, in the sciences: Using the neuron to construct a starting framework. Minds and Machines, 23(3), 309–337.

    Article  Google Scholar 

  • Glennan, S. S. (1996). Mechanisms and the nature of causation. Erkenntnis, 44(1), 49–71.

    Article  Google Scholar 

  • Halpern, J.Y., Hitchcock, C. (2011). Actual causation and the art of modeling. arXiv:1106.2652.

  • Harbecke, J. (2013). The role of supervenience and constitution in neuroscientific research. Synthese, 191, 1–19.

    Google Scholar 

  • Harinen, T. (unpublished). Towards an account of scientific constitution. Unpublished manuscript.

  • Hausman, D. M., & Woodward, J. (1999). Independence, invariance and the causal markov condition. The British Journal for the Philosophy of Science, 50(4), 521–583.

    Article  Google Scholar 

  • Kim, J. (1984). Concepts of supervenience. Philosophy and Phenomenological Research, 45(2), 153–176.

    Article  Google Scholar 

  • Kim, J. (1988). Supervenience for multiple domains. Philosophical Topics, 16(1), 129–150.

    Article  Google Scholar 

  • Kim, J. (2000). Mind in a physical world: An essay on the mind-body problem and mental causation. Cambridge: MIT press.

    Google Scholar 

  • Leuridan, B. (2012). Three problems for the mutual manipulability account of constitutive relevance in mechanisms. The British Journal for the Philosophy of Science, 63(2), 399–427.

    Article  Google Scholar 

  • Lewis, D. K. (1973). Counterfactuals. Oxford: Blackwell Publishers.

    Google Scholar 

  • Machamer, P., Darden, L., & Craver, C. F. (2000). Thinking about mechanisms. Philosophy of science, 67, 1–25.

    Article  Google Scholar 

  • McKitrick, J. (2003). A case for extrinsic dispositions. Australasian Journal of Philosophy, 81(2), 155–174.

    Article  Google Scholar 

  • McLaughlin, B., & Bennett, K. (2014). Supervenience. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy. http://plato.stanford.edu/archives/spr2014/entries/supervenience/.

  • Menzies, P. (2012). The causal structure of mechanisms. Studies in History and Philosophy of Science Part C, 43(4), 796–805.

    Article  Google Scholar 

  • Mumford, S. (1994). Dispositions, supervenience and reduction. The Philosophical Quarterly, 44(177), 419–438.

    Article  Google Scholar 

  • Polger, T. W. (2010). Mechanisms and explanatory realization relations. Synthese, 177(2), 193–212.

    Article  Google Scholar 

  • Schindler, S. (2013). Mechanistic explanation: asymmetry lost. In V. Karakostas & D. Dieks (Eds.), EPSA11 perspectives and foundational problems in philosophy of science. The European Philosophy of Science Association Proceedings (pp. 81–91). Springer.

  • Shapiro, L. A. (2012). Mental manipulations and the problem of causal exclusion. Australasian Journal of Philosophy, 90(3), 507–524.

    Article  Google Scholar 

  • Shapiro, L., & Sober, E. (2007). Epiphenomenalism - The do’s and the don’ts. In P. Machamer & G. Wolters (Eds.), Thinking about Causes (pp. 235–264). Pittsburgh: University of Pittsburgh Press.

  • Soom, P. (2012). Mechanisms, determination and the metaphysics of neuroscience. Studies in History and Philosophy of Science Part C, 43(3), 655–664.

    Article  Google Scholar 

  • Woodward, J. (2005). Making things happen: A theory of causal explanation. New York: Oxford University Press.

    Google Scholar 

  • Woodward, J. (2014). Interventionism and causal exclusion. Philosophy and Phenomenological Research. doi:10.1111/phpr.12095.

  • Woodward, J., & Hitchcock, C. (2003). Explanatory generalizations, part i: A counterfactual account. Nous, 37(1), 1–24.

    Article  Google Scholar 

Download references

Acknowledgments

I thank Carl Craver, Eleanor Knox, David Papineau and two anonymous reviewers for valuable comments on previous versions of the manuscript. I’m also grateful to all participants and organizers of the ‘Mind and Life: Mechanistic and Topological Perspectives’ conference in Belgrade. The research that went into this paper was supported by Kone Foundation.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Totte Harinen.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Harinen, T. Mutual manipulability and causal inbetweenness. Synthese 195, 35–54 (2018). https://doi.org/10.1007/s11229-014-0564-5

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11229-014-0564-5

Keywords

Navigation