A Defence of Functional Kinds: Multiple Realisability and Explanatory Counterfactuals

ABSTRACT

In this paper, I defend an updated account of functional kinds, initially presented by Daniel Weiskopf, from the criticism that functional kinds will not qualify as scientific kinds. An important part of Weiskopf’s account is that functional kinds are multiply realisable. The criticisms I consider avoid discussion of multiple realisability. Instead, it is argued that functional kinds carry inferior counterfactual profiles when compared to other accounts of kinds. I respond to this charge by arguing that this criticism fails to take into consideration the role that multiple realisability can play in providing important explanatory counterfactuals. I do so by highlighting some points made by Lauren Ross that highlight where multiple realisability is explanatorily pertinent. I then argue that the criticisms of Weiskopf’s account fail to establish the explanatory inferiority of functional kinds.

Notes

1 Weiskopf allows that some may simply be difficult to reduce, rather than irreducible in principle. I will not consider this deviation here as all that is important there are some models that cannot be reducible.

2 Buckner uses an example of backpropagation because he does not think that FELs meet Weiskopf’s requirement of being applied to several, distinct, targets. This concern does not effect my arguments so I focus on FELs.

3 See Hummel and Biederman (1992, 484–485) for a discussion.

4 Weiskopf does not commit himself to all functional kinds being multiply realisable. However, Weiskopf is committed to their being some multiply realisable psychological kinds (2011a), which is all that is needed to get this defence off the ground.

5 Older accounts of functionalist philosophies will replace (3) and (4) in Weiskopf’s account with (3’) a nomological approach to induction and explanation, and, (4’) a nomological account of kinds (Buckner 2015, 3920).

6 This is to say that the criticism being levied is not one that relies on the ‘perfect model model’ of representation, where a model is only explanatory if it includes no misrepresentations (Teller 2001). Rather, given that the model is meant to provide an explanation of some phenomenon, the concern is that the misrepresentations should not be found in what is meant to be explanatory.

7 I am assuming here that this is enough to count as multiple realisation. There are different accounts of multiple realisation, some of which would not consider this multiple realisation. See Shapiro (2004) and Sullivan (2008) for a discussion of this.

8 Thank you to a reviewer for pressing me on this point.

9 One might respond that we do not have to take the mechanistic explanation, in relation to a heterogeneous set, to be making this sort of commitment to particular components. Just as the functional explanation might include ‘components’ like FELs, we might take the components of a mechanistic explanation to also simply be providing functional capacities. This, however, is not a mechanistic explanation as it is often outlined (Kaplan and Craver 2011).

10 Note that a response cannot be that mechanisms and functional kinds capture different capacities as this would make the ceteris paribus clause of Buckner’s comparison moot.

11 While these criticisms are rarely presented in terms of multiple realisability, the implication from these may undercut the multiple realisability that functionalists often help themselves to.

12 Not all accounts of multiple realisability come from functionalists. See Piccinini and Maley (2014); and Rosenberg (2015) for discussions of multiple realisability within a mechanistic framework.