What topic continuity problem?

ABSTRACT

A common objection to the very idea of conceptual engineering is the topic continuity problem: whenever one tries to ‘reengineer’ a concept, one only shifts attention away from one concept to another. Put differently, there is no such thing as conceptual revision: there's only conceptual replacement. Here, I show that topic continuity is compatible with conceptual replacement. Whether the topic is preserved in an act of conceptual replacement simply depends on what is being replaced (a conceptual tool or a conceptual role) and what the topic under discussion is. Thus, the topic continuity problem only arises from a failure to specify these two things.

KEYWORDS:

• Conceptual engineering
• conceptual replacement
• topics
• conceptual roles
• conceptual tools

Acknowledgments

Many thanks to Rachel Rudolph, David Plunkett, an anonymous referee, and the editor for their feedback on this paper.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 Also known as the ‘changing the subject objection’ or ‘Strawson's challenge’.

2 These are my best attempt to provide explicit premise-conclusion formulations of these problems, which I have yet to find in the literature. They are admittedly somewhat unclear, however. For example, premise 2 does not state what the topic being changed by conceptual replacement is. (Indeed, I think this unclarity is precisely where the arguments falter; see §5.) Still, these formulations are broad enough to encompass the many variants of the problem from the literature, including those articulated by Cappelen (2018, 101–102), and to sort the many different kinds of responses that have been presented.

3 This view is clearly related to what is called functonalism in the philosophy of mind, which holds that what makes something a mental state of a particular sort (belief, desire, etc.) is its functional role. See Levin 2018 for an overview.

4 For ease of exposition, I'm using ‘the meaning of a word’ to refer to semantic meaning (what an expression means in a language or linguistic community), rather than speaker meaning (what a particular speaker means by an expression). This is what Cappelen seems to have in mind. See Vermeulen (2018), Deutsch (2020), Koch (2021a), and Pinder (2021) for more on this distinction (which is originally due to Grice 1968) in the context of conceptual engineering.

5 I set aside Cappelen's (2018) doubts over the very idea of a concept's function. See Simion and Kelp (2020) and Thomasson (2020) for responses. Cappelen mainly objects to the idea that concepts have ‘central’ or ‘proper’ functions (Haslanger 2000; Thomasson 2020). My notion of a conceptual role does not assume (or deny) that the conceptual role filled by a conceptual tool is that tool's central or proper function in this sense.

6 Prinzing distinguishes between two senses of what a concept is ‘for’, viz., the point of the concept in a particular case vs. the point of a concept in the first place (cf. the distinction between etiological and design functions; Simion and Kelp 2020). He's clear that he has in mind the latter (868).

7 Formally, a partition on W is a set $\mathrm{\Pi }\subseteq \mathrm{\wp }W$ such that (i) $\varnothing \notin \mathrm{\Pi }$, (ii) $\bigcup \mathrm{\Pi }=W$, and (iii) $A\cap B=\varnothing$ for all distinct $A,B\in \mathrm{\Pi }$. A ‘cell’ of Π is just a member of Π. We can also think of Π in terms of the equivalence relation ${\sim }_{\mathrm{\Pi }}$, where $x{\sim }_{\mathrm{\Pi }}y$ iff there is some $A\in \mathrm{\Pi }$ such that $x,y\in A$.

8 See Parry (1968), Perry (1989), Railton (1993), Yablo (2014), Fine (2016), Cappelen (2018), Hawke (2018), and McPherson and Plunkett (2021) for alternative approaches.

9 See, e.g. Hamblin (1958), Hamblin (1973), Karttunen (1977), Groenendijk and Stokhof (1984), Groenendijk (1999), Groenendijk (2009), and Ciardelli, Groenendijk, and Roelofsen (2018). See Cross and Roelofsen (2018) for an overview.

10 Another option is to treat practical questions as normative questions, i.e. questions about what we should do (though see Balcerak Jackson 2019; Risberg 2022). Yet another option (one that I prefer) is to treat practical questions as sui generis, in that their answers are not analyzed in terms of sets of worlds but something else altogether (e.g. plans).

11 We could also talk about conceptual tool creation, i.e. introducing a tool to a fill a previously unfilled role, or conceptual tool destruction, i.e. removing a tool from filling a role without replacement (cf. Cappelen 2018; Simion and Kelp 2020). Similarly for conceptual role creation and destruction. The solution in §5 applies equally to conceptual creation and destruction.

12 Since there is a one-to-one correspondence between expressions e and properties of the form being the meaning of e, we could have equivalently taken $r_e$ to just be e and interpreted the filling relation as the relation being the meaning of.

13 The question of whether we have power to change which tools fill which roles should be distinguished from the question of whether filling is an absolute or relative matter, i.e. whether tools fill roles simpliciter or whether individuals or communities fill roles with tools. In one direction, we may have the power to change which tools fill which roles even if filling is absolute. For example, if roles are properties and filling is instantiation, then filling is absolute. But this is compatible with us having the power to change which tools instantiate which role-properties. In the other direction, it is possible for us to lack the power to change which tools fill which roles even if filling is a relative matter. For example, it may be that whether a tool fills a role depends on the origins of a society, which cannot be changed.

14 With that said, even if engineering concepts is hard, it may still be worth pursuing (Cappelen 2018; Koch 2021a). Moreover, even if it's impossible, it may still be possible (and fruitful) to engage in other conceptual activities such as conceptual innovation (Simion and Kelp 2020) or conceptual exploration (Rudolph 2021).

15 Here, I'm setting aside issues that arise from Kaplan's paradox (Kaplan 1995). If intensions are functions from worlds to extensions, then there are strictly more intensions than there are worlds by Cantor's theorem. But if there's a world for every intension used to assign to ‘planet’, then there need to be at least as many worlds are there are intensions. If this undermines the Lewisian model of topics, so be it: the model is inessential to my proposed solution to the topic continuity problem.

16 Nado does say, ‘The needed functional continuity is, moreover, very flexible–functions can be rejected, traded off, split, combined, reshuffled’, suggesting she would allow for some differences in function. But this only delays the problem of ‘demarcating the limits of permissible revision’: the question then becomes how much functional discontinuity is compatible with conceptual engineering? My proposal can be understood as an answer to this question: the amount of functional discontinuity compatible with an engineering project is simply demarcated by the topic under discussion.

17 This illustrates how my proposal differs from Koch's (2021b) explanatory eliminitativism about topics. Koch argues topics play no explanatory role in conceptual engineering, whereas I maintain they explain what kinds of engineering proposals address the conceptual task in question. Still, I agree with the spirit of Koch's diagnosis: conceptual engineers need not invest heavily in demarcating ‘the limits of revision’. They simply need to be clear about the terms of the debate: the limits of revision are demarcated by the topic of discussion.

18 Nado (2021, S1519) gives a similar example of choosing between different biological taxonomies. Nado argues that cases like this really preserve the function of sets of concepts. While this could be a way of thinking about such cases, it is not enough to defend the claim that role continuity is necessary for topic continuity. After all, the topic may be to find a role for a particular set of conceptual tools to fill (e.g. the IAU could decide to keep the old classification system but to give them different labels or use them for different purposes).

19 Here, I am not taking a stand on the exact mechanism that gives rises to these different readings. For different proposals, see Belleri (2017), Thomasson (2017), Kocurek, Jerzak, and Rudolph (2020), and Mankowitz (2021).

20 This distinction is similar to Belleri's (2021) distinction between conservative inquiries, which require preserving the current meaning of ‘F’, and semantically progressive inquiries, which do not. While Belleri views this as a distinction between two forms of inquiry over a single question, I construe it more directly as a distinction between two readings of a question. Thus, semantically conservative inquiries are simply inquiries into factual questions while semantically progressive inquiries are inquiries into interpretative questions.

21 While I've been construing the ‘interpretative’ reading of the question ‘What is a planet?’ in practical terms, some might prefer to construe them in normative terms. For example, for (Plunkett and Sundell 2013), what I'm calling the ‘interpretative’ reading amounts to asking how ‘planet’ should be interpreted or how best to interpret it–or, in terms of tools and roles, which conceptual tools should (or best) fill which conceptual roles. Here, it does not matter for my purposes whether we construe this reading in practical or normative terms. Arguably, in most contexts, the two go hand-in-hand. What's important is that there is a reading of ‘What is F?’ questions that can be understood in terms of conceptual tools and conceptual roles.

22 Compare this to Chalmers's (2011) claim that disputes over ‘What is F?’ questions tend to be verbal. Indeed, on the current proposal, philosophical disputes over ‘What is F?’ questions will be verbal in the sense that they are disputes over how to interpret ‘F’. Unlike Chalmers, however, I do not conclude from this that such disputes are pointless or unimportant for the purposes of philosophical inquiry (cf. Plunkett 2015; Belleri 2017; Thomasson 2017; Knoll 2020).