Complexity-based Theories of Emergence: Criticisms and Constraints

Abstract

In recent years, many philosophers of science have attempted to articulate a theory of non-epistemic emergence that is compatible with mechanistic explanation and incompatible with reductionism. The 2005 account of Fred C. Boogerd et al. has been particularly influential. They argued that a systemic property was emergent if it could not be predicted from the behaviour of less complex systems. Here, I argue that Boogerd et al.'s attempt to ground emergence in complexity guarantees that we will see emergence, but at the cost of rendering it either trivial or epistemic. There are three basic problems. First, neither the measures of complexity explicitly mentioned by Boogerd et al. nor the most popular measures in the literature can do the practical and theoretical work that they assign to complexity. Second, I argue that while the success of their view depends on restricting the base of information available to the reductionist, this cannot be done in a way that is metaphysically neutral with respect to emergence. Thus, their account renders emergence trivial. Third, I argue that grounding emergence in complexity can support only epistemic emergence. I conclude by considering the methodological import of their account.

Acknowledgements

For valuable comments and discussion, I would like to thank Antony Aumann, Irina Mikhalevich, Elisabeth Lloyd, Tim O'Connor, Carlos Zednik, Andrew McAninch, Cameron Buckner, Dominique Chu, and many other teachers, friends, and colleagues. I am especially grateful to Colin Allen for patiently working through many iterations of this paper with me. I also thank James W. McAllister and two anonymous referees for this journal for their insights and suggestions. Thanks also to Myles and Peg Brand and to the participants and organizers of ‘Philosophical Foundations for Systems Biology’ at the University of Oslo.

Notes

[1] For a representative sample, see Bunge (1981), Beckermann (1992), Sperry (1992), Bechtel and Richardson (1993), Newman (1996), Bedau (1997, 2008), Stephan (1997, 2002, 2006), Silberstein and McGeever (1999), Gillett (2002), Boogerd et al. (2005), Korn (2005), Francescotti (2007), Grantham (2007), Richardson and Stephan (2007), Hovda (2008), Huneman and Humphreys (2008), Symons (2008), Bich (2012), Huneman (2012), and Hendry (forthcoming).

[2] See Bedau (1997, 2002), Grantham (2007), Hovda (2008), Symons (2008), Bich (2012), and Mitchell (2012). For alternative accounts of ontological emergence, see Bunge (1981), Sperry (1992), Newman (1996), Stephan (1997, 2002, 2006), Gillett (2002), Francescotti (2007), Huneman (2012), and Hendry (forthcoming).

[3] Many have argued that emergence and reduction are compatible (Wimsatt 1997; Butterfield 2011a, 2011b). I take no issue with such claims.

[4] For an incisive discussion of this requirement, see Humphreys (2008). He remarks:

What I call the sparse perspective on emergence is probably the dominant perspective in the philosophical literature on emergence. The sparse perspective holds that emergence, if it exists at all, is a rare phenomenon. Indeed, the sparse perspective often seems to be used as an informal touchstone for accounts of emergence: if the account makes emergence too common, that is a reason to think the account is defective (Humphreys 2008, 593).

[5] It also distinguishes this group's account from that of Stephan (2002), in which he offers diachronic unpredictability as a condition that distinguishes strong and weak emergence.

[6] For example, downward causation would violate the vertical condition because it involves systemic activity that cannot be explained at lower levels. Such cases are synchronically emergent and resist mechanistic explanation.

[7] Boogerd et al. appear to use ‘prediction’ and ‘deduction’ interchangeably. Where possible, I have restricted my discussion to prediction, for two reasons. First, prediction and deduction are not obviously equivalent, and this terminological choice should be motivated but appears not to be. I thank an anonymous reviewer for drawing my attention to this point. Second, Boogerd et al.'s use of ‘deduction’ in this context presumes, of course, that the reductionist is committed to a view of explanation on which explanation requires logical deduction. I do not wish to defend such a view. It should also be noted that there is also dispute concerning whether Broad's original formulation required a formal derivation (Humphreys 2008) or something weaker than formal derivation (Richardson and Stephan 2007).

[8] Bob Richardson (personal communication) has suggested that, at the time of their writing, this is close to what he had in mind. Of course, what constitutes a ‘part’ raises serious metaphysical difficulties. While I recognize the importance of these questions, and the additional problems it may present for this measure of complexity, such issues are outside the scope of the present work.

[9] This, of course, is an oversimplification. One encounters numerous confounds when attempting to perform this type of experiment. I will return to this point shortly.

[10] The SHH gene and the protein it produces are referred to by the same name. Following standard convention, I will try to avoid confusion by using SHH to refer to the gene and SHH to refer to the protein.

[11] I thank an anonymous reviewer for this suggestion.

[12] Here, I am assuming scientific realism. I do not take myself to be responsible for defending this assumption here. In doing so, I am following Machamer, Darden, and Craver (2000), as well as many other mechanists.