“Conceptual engineering” is a label for the project of investigating, evaluating, and, when necessary, repairing, replacing, or augmenting the elements of our own conceptual toolkit. In Fixing Language (2018), Herman Cappelen develops a theoretical framework for understanding this kind of project. He defends conceptual engineering from certain key objections and offers a specific, and in some ways idiosyncratic, account of what the project amounts to. For Cappelen, the joint projects of defending conceptual engineering, on the one hand, and explaining it, on the other, are closely intertwined. While some aspects of his framework stand more or less on their own—Cappelen’s externalist metasemantics, for example—others are bound up with his reply to a set of foundational objections to the whole endeavor. Those objections are what Cappelen calls “the Strawsonian challenges.” They are all versions of this: If I ask you a question about the F’s, and you give me an answer that’s not about the F’s but rather about the G’s, then you haven’t answered my question. You have changed the subject. Reengineering or replacing our concepts changes which things in the world we’re talking about. What would be the point of that? It was the first set of things we were supposed to be investigating or making choices about. If the challenge lands, then conceptual engineering is a misguided endeavor. It is a recipe for talking past our interlocutors and failing to answer the questions that motivated our inquiry and practices in the first place.
In this paper I describe Cappelen’s understanding of the Strawsonian challenge, his “coarseness of topic” response to that challenge, and how that response works to shape his positive account of conceptual engineering. I argue that Cappelen’s response succeeds in reply to one understanding of the Strawsonian challenge—on which it is motivated by ordinary judgments of samesaying and continuity of topic—but that it fails as a response to another version—on which a parallel objection is motivated by philosophical considerations, and is stated in a theoretical register. The absence of a response to this second challenge reveals an important gap in Cappelen’s framework. Ultimately, I argue, Cappelen’s conflating of the two types of challenge boils down to a failure in, of all things, conceptual engineering. In putting our pretheoretic notion of topic to work in addressing theoretical worries framed in technical terms—a use for which it wasn’t intended, and to which it is not particularly well suited—Cappelen refrains from engaging in precisely the kind of interrogation of our representational devices for which the rest of his book serves as a manifesto.Footnote 1
1. Changing the subject: continuity of topic and saying what others said
The general thought behind conceptual engineering is something like this: On the one hand, we could engage in practical or theoretical inquiry by formulating questions, and then working to find out what the answers to those questions are. On the other hand, we could look at the concepts with which we formulate those questions in the first place. We could ask if those are, in fact, the concepts we ought to be using. Perhaps we’ll find that given the kind of inquiry or practice we’re engaged in, our concepts are defective in crucial respects—that we’re asking the questions in the wrong terms. If our concepts are defective in some way, we could respond by changing them, or replacing them, finding new concepts to work alongside them, or getting rid of them altogether.
If this rough sketch of a project looks compelling to you, then further questions immediately present themselves. What would it mean for a concept to be defective? (Cappelen gives examples of concepts that are purported to be defective logically, cognitively, scientifically, and morally/politically.) What exactly are the things being engineered? (Cappelen ultimately rejects talk of engineering “concepts” altogether, preferring to speak purely in terms of changing the intensions and extensions of expressions.Footnote 2 I’m happy to follow him in that respect here.) How often does conceptual engineering actually happen? (Cappelen, along with many in the conceptual engineering literature, suggests that this is what we’re doing, both within and outside of philosophy, much of the time.) How difficult is it? (Cappelen would say “very.”) How do we know when we’ve succeeded? (Cappelen would say we might not.)
All these questions have a range of plausible answers. But whatever answers you give them, you are bound to run into a particular kind of objection before long. Cappelen traces the objection to Strawson’s response to Carnap’s notion of “explication.” The passage from Strawson he cites is worth reproducing here.
To offer formal explanations of key terms of scientific theories to one who seeks philosophical illumination of essential concepts of non-scientific discourse, is to do something utterly irrelevant—is a sheer misunderstanding, like offering a textbook on physiology to someone who says (with a sigh) that he wished he understood the workings of the human heart. … Typical philosophical problems about the concepts used in nonscientific discourse cannot be solved by laying down the rules of exact and fruitful concepts in science. To do this last is not to solve the typical philosophical problem, but to change the subject. (1963, 505)
Cappelen goes on to cite other more recent philosophers working in a range of areas who express worries in a similar vein: Mark Richard (Reference Richard2019) on Haslanger’s ameliorative analyses of “race” and “gender”: “For that matter, doesn’t [Haslanger’s revisionary project] cross the line between conceptual therapy and stipulative rebranding?” Peter Ludlow (Reference Ludlow, Preyer and Peter2005) on anticontextualists in epistemology: “Any investigation into the nature of knowledge which did not conform to some significant degree with the semantics of the term ‘knows’ would simply be missing the point.” Peter Railton (Reference Railton1989) on his own revisionary naturalism in metaethics: “[If a revisionary metaethical naturalist] wishes to make his case compelling, he must show that his account of a person’s good is a rather clear case of tolerable revision, at worst.” Quoted first among these by Cappelen, and summing up the various concerns most pointedly, is Haslanger herself: “Revisionary projects are in danger of providing answers to questions that weren’t being asked” (2000, 34; emphasis added).
Cappelen draws out from these passages a number of distinct concerns, but I’ll focus on two of them. The first is what I’ll call the “continuity of topic” objection.Footnote 3 The second is the “reporting” objection.
The first objection—continuity of topic—concerns the idea that if we change the intension and extension of the expressions we use in some inquiry, then we are no longer talking about the same thing. Here Cappelen summarizes the continuity concern with conceptual revision:
Even if the revisions succeed, they do not provide us with a better way to talk about what we were talking about; they simply change the topic… . The objection is that the answers employing terms with new extensions fail to answer the original questions. These answers concern something new—not what we were originally talking about when we used the [original expressions]. We have the illusion of an answer, but it’s a purely verbal illusion. There’s a lack of continuity of inquiry; the old questions are not being answered. (100–102)Footnote 4
In a similar vein, Cappelen discusses what he calls the “verbal disputes” objection. This is essentially the same worry as that described above but seen through the lens of disagreement in particular. (Hence, I’ll treat it as one part of the continuity of topic objection.) If we change the extension and intension of the expressions central to some debate, then won’t that debate devolve into merely verbal disputes—participants simply arguing past one another—as opposed to a dispute expressing a substantive disagreement about the topic under investigation?
Worries about changing the topic and about verbal disputes are thus two sides of the same coin. It seems to us that when we engage in sustained inquiry into some topic—when we investigate or make choices about it, when we change our beliefs, or when we debate it—we have, in the typical case, the ability to be talking about that same topic over time, across theories, contexts, and viewpoints, even as we change our own beliefs or express disagreement with others. Despite these differences and changes, there remains a single topic of our discussion—a sort of stable communicative ground. The project of conceptual engineering would seem to threaten this crucial aspect of our inquiry and practice. If instead of discussing and debating a single topic over time and across speakers, we instead go and change the meanings of the terms central to those discussions and debates, then we have lost that stable ground.
That is the worry at the heart of the continuity of topic objection to conceptual engineering. But the second objection—the reporting objection—is different. Cappelen asks how, once the project of conceptual engineering has begun, we should go about reporting the speech of people who employ the engineered terms with their earlier meanings. Using the example of a changed meaning for the word ‘belief’ (in the spirit of Clark and Chalmers), Cappelen suggests that in reporting the speech of previous users:
If we use the word ‘belief’ with its new meaning, then it looks like we’re misquoting past speakers. So the right thing to do would be to just quote them—to say that they used the word ‘belief,’ but not to use that word to say what they said. (102)
That’s what it seems we would have to do if conceptual engineers succeeded in changing the meaning of the relevant term. If conceptual engineers actually change the extension and intension of a term, then much of our innocuous-seeming indirect speech reporting would simply be false. If the meaning of our word “belief” has changed, then earlier users of that word weren’t saying anything about beliefs, and to describe them as doing so with an indirect speech report is simply mistaken. To get things right, we would have to switch to direct quotation. And yet, it’s hard to imagine the circumstances under which we’d actually feel compelled to change our practice of speech reporting like this.
How exactly this objection plays out depends on whether you think we should start doing conceptual engineering or whether we’ve actually been doing it all along. (Conceptual engineering advocates can differ on this point.) If you think that conceptual engineering is something that we should start doing, then this observation suggests that you’ll have to be prepared to radically alter our practices of reporting the speech and attitudes of others in any domain where our engineering could prove successful. Breaking such a useful practice is not an appealing prospect. If, on the other hand, you think that conceptual engineering is something that we’ve been doing all along, then you are essentially forced to a kind of error theory about much of the speech and attitude reporting that’s gone on in the domain where engineering has taken place. If conceptual engineering has been taking place all along, then much of our indirect speech reporting in that domain is wrong. But our indirect speech reporting practices seem to be just fine. So conceptual engineering can’t have been taking place.
2. Cappelen’s response
2.a Topics are coarse grained
How does Cappelen respond to this pair of worries? He summarizes it himself in the following slogan: “Sameness of topic doesn’t track sameness of extension” (109). What Cappelen means by this is that it is possible for two speakers to use the same expression with different extensions and, nevertheless, for those speakers to be talking about the same topic.
Capellen frames the subsequent discussion in terms of extensions in particular. But earlier in the book, he makes it clear that when he talks about a change in extension, what he means is a change in extension that results from a change in intension (62). The reason for this proviso is that extensions can change over time simply because the world changes. The extension of ‘salad’ changes over time because salads get made and eaten. So a use of the word ‘salad’ yesterday would have a different extension than a use of that word today even supposing that, in the relevant sense, the “meaning” has remained entirely the same. So sameness of topic over time couldn’t possibly require identity of extension, even as a first pass. What really matters here is identity of intension. If the intension of ‘salad’ changes, then what it takes to be the kind of thing ‘salad’ is true of has changed. The extension will change as well, of course, but not just because old salads have been eaten and new salads have been made, but rather (and in addition) because the criteria for satisfying the predicate have changed. It’s those changes in criteria—in the expression’s intension—that interests the conceptual engineer, and that the Strawsonian objector worries about. For this reason, although Capellen often states his “coarse-grainedness” response in terms of extensions, I’ll focus on the intensions of the relevant expressions. The slogan for purposes of discussion here is thus: Sameness of topic doesn’t track sameness of intension. Footnote 5
Why should we think this slogan is true, and how does the observation it describes serve as a response to the continuity of topic and the reporting objections? Start with why we should think it’s true. Here, Cappelen—as he has in earlier workFootnote 6—draws our attention to the fact that in the case of many uncontroversially context-sensitive expressions, we seem happy to report two speakers as having “said the same thing” in situations where it’s quite obvious that the speakers could not have been using those expressions with the same intension.
Consider, for example, two speakers who describe, in different contexts, a particular political event as “significant.” The character of a word is a function from a context of utterance to an intension. And in the case of a word like ‘significant,’ the resulting intension (and of course, correspondingly, the extension) depends on the context in especially complicated ways. ‘Significant’ is a relative gradable adjective. So that means that the context will need to supply a threshold: How high on the scale of significance does something have to be to count as “significant”? But, clearly, the nature of that scale is itself in need of contextual resolution. Significance to what or to whom? What factors matter in determining significance? How are those factors combined and weighed against each other in creating a unified scale of significance? The answers to these questions will vary greatly and in subtle ways from context to context. Of course, the character of the word ‘significant’—the function from contexts to intensions—might be exactly the same for the two speakers. But because of the complex way in which the context and character conspire to fix an intension, identity of character is no guarantee of identity of intension, even in vaguely similar contexts. Plausibly, the chance that speakers in two different contexts ultimately pick out the same kinds of objects or events with the word ‘significant’—that they use that word with precisely the same intension—is close to zero.
And yet, it’s easy to imagine a third party, aware of the two utterances, reporting those two speakers as having said the same thing.
-
(1) Speaker A and Speaker B both said that the event is significant.
-
(2) Speakers A and B agree that the event is significant.
-
(3) Speakers A and B said the same thing about that event—namely, that it’s significant.
All these reports use the same expressions that the speakers themselves used in their original utterances—they are disquotational speech reports. And the disquotational speech reports in (1) through (3) sound like perfectly reasonable descriptions of this state of affairs. So what, exactly, is it that both Speaker A and Speaker B said? That the event is significant! Of course, it’s possible to cook up a pair of contexts where the scales or the standards for significance differ so greatly between the contexts of Speakers A and B that some of these reports would strike us as fishy. That’s no problem for Cappelen. What matters is that, in a huge number of cases, this kind of reporting doesn’t seem fishy in the least. These reports strike us as unproblematic, as an extremely useful way to report the facts. They strike us as true.Footnote 7
What Cappelen points out, in bringing this observation back to the issue of topic continuity, is that the topic of a conversation or an inquiry is a matter of what it is you’re talking about. So if two people are saying the same thing, it follows trivially that they’re talking about the same topic. What is the topic Speakers A and B are talking about? It’s simple: the political event, and whether it’s significant. In this case, they happen to agree—it is significant. But note that if they disagreed—if one of them said “the event is significant,” and the other said “the event is not significant”—it would be the same story. A and B were talking about the same thing—whether the event was significant—and in the disagreement case they just happen to have made opposing claims about it.
The fact that we can report these two speakers as talking about the same thing in this, our own context, just makes Cappelen’s point that much more vivid. Our context is no more likely to overlap perfectly with either of the speakers’ contexts in the intension of ‘significant’ than those contexts are likely to overlap each other. And yet, knowing that, in all likelihood, each of the three uses of ‘significant’ has different intensions, these reports still strike us as perfectly acceptable. The only conclusion is that the notion of “topic” at work in our claims about whether speakers say the same thing (whether that are “samesaying”), or whether they’re talking about the same thing, etc., is sufficiently coarse-grained as to permit not just variation in extension but also variation in intension.Footnote 8
So how does this observation serve as a response to the continuity of topic and the saying what others said objections? The application is straightforward. If someone accuses the conceptual engineer of changing the meanings of the expressions that matter, and thus switching topics, the conceptual engineer can respond that, yes, they changed the intension of certain expressions, but it simply doesn’t follow that the topic itself was changed. To think that it does follow is to fail to recognize that the notion of topic never required identity of intension in the first place. If two people who mean different things by ‘significant’—who use it with different intensions—can nevertheless both be addressing the topic of whether some event was significant, then two people who mean different things by ‘gender,’ or ‘race,’ or ‘marriage,’ or ‘salad’ can both be investigating or debating the nature of gender or race or marriage or salad.
The application to the reporting objection is even more direct. When we report others’ speech—even when we do so disquotationally—it’s simply not the case that what we’re doing is reporting them as having uttered expressions with the same intensions as the expressions we use in our report. When I report Speakers A and B from above as having both said the same thing—namely, that the event is significant—my report is true. So the worry that the conceptual engineer would be committed either to changing our practice of speech reporting, or to an error theory about it, is misguided. It misunderstands what we report when we make those speech reports. We were never reporting that the two speakers used their words with exactly the same intension. So a theory on which they didn’t is not necessarily a theory on which our reports are mistaken.
2.b The Contestation Theory
Cappelen’s view that intensions can shift while topics remain the same immediately raises the following question: “How much can they shift?” Cappelen’s response is what he calls the “Contestation Theory.” To the question just posed, Cappelen responds that any descriptive answer a particular theorist offers is itself up for debate for the conceptual engineer. Depending on our interests in a particular context, a certain shift in intension might be entirely acceptable as a topic-preserving proposal. Or, on the other hand, it might go altogether too far and count as a change in topic. In other words, it’s not the kind of question that can be decided ahead of time or in any way other than on a case by case basis. While any number of theorists have proposed accounts of what it takes for a certain revision to, in fact, count as topic preserving, Cappelen insists that we view such accounts as at best normative proposals—proposals which might be appropriate for some debates and inappropriate to others. Even if a certain kind of shift has historically been treated by everyone as too radical to maintain topic continuity, we might, in the future, decide that there is some thread linking the usages that’s continuous enough and, in context, important enough, to count the shift as topic preserving. Conversely, a shift that has historically been treated as minor enough to preserve topic might be significant enough in the respects that matter to justify a verdict of topic change in future cases. Just like the choice of meaning itself, the requirements for topic preservation in the face of meaning change are up to us.
2.c The “worldliness” of conceptual engineering
As I suggested in the introduction, Cappelen’s response to the Strawsonian objections plays an important role in shaping his positive view of conceptual engineering. Soon after the chapters replying to the Strawsonian objections, Cappelen poses the question of exactly what gets engineered by the conceptual engineer (137). He considers three options: (1) the conceptual engineer engineers concepts, (2) the conceptual engineer engineers words and their meanings, and (3) the conceptual engineer engineers the world itself.
Cappelen rejects the view in option (1). As noted, he rejects talk of concepts (in this context) altogether. As for option (2), Cappelen obviously thinks there is some truth to it. When conceptual engineering succeeds, it does result in a change in the intensions of certain words. But Cappelen thinks this is not the only, or even the most interesting or significant, way to describe what the conceptual engineer achieves. What Cappelen ultimately endorses is option (3): the conceptual engineer engineers the world itself. “So construed,” he says, “the result of conceptual engineering can be described as an object-level change: we’re changing what gender, freedom, salad, marriage, etc. are” (137; emphasis in original). This is a strong claim. The suggestion that salads themselves—or freedom, or truth, or belief, or persons, or any other target of conceptual engineering—can change as a result of our choices about how to talk could surprise even an impassioned advocate of conceptual engineering. It’s one thing to argue for the practical or theoretical significance of our choices about how to carve up the world conceptually or how to use our words in certain contexts. It’s another to say that those choices effect object-level changes on the world. How can Cappelen defend the stronger claim?
What Cappelen relies on here is precisely his discussion of the coarse grainedness of topics. He considers a situation where the meaning of the word ‘family’ has changed between time t and time t*, and the following description of that situation:
(4) What families are has changed.
The use of ‘family’ in (4) could have the intension it did at t, or the intension it did at t*, or some third intension it acquired by the time of the utterance of (4). And Cappelen concedes that no matter which of those we choose, the utterance of (4) will semantically express a false proposition. After all, the things in the extension of the word ‘family’ as it is used in (4) haven’t changed, or at least not as a result of the meaning change the speaker of (4) is referring to. But Cappelen thinks that utterances express a wide range of expressions over and above what they semantically express. And the proposition expressed by an utterance of (4) that’s of interest to Cappelen is the one in which ‘family’ serves to designate the kind of coarse-grained topic that has remained fixed even as the meaning of the word ‘family’ has changed. This proposition, also expressed by the utterance of (4) (just not semantically expressed), is true. And it is the proposition that Cappelen identifies as best capturing what the conceptual engineer is up to. In this case, changing what families are.
And yet, despite advocating for this option as the best characterization of the conceptual engineer’s project, Cappelen is careful to point out that the suggestion is to be taken in a “metaphysically lightweight” sense. As he puts it,
The underlying mechanisms [of change] involve expressions changing their extensions and intensions and a recognition that these changes are topic-preserving. The result is that we can say truly that, e.g., families have changed (and we can see that change as a result of conceptual engineering). [But] [t]he families that end up existing at some time/world pair is the result of familiar facts about people creating bonds in various complicated, non-spooky ways. There are true object-level ways that can be used to describe the results of conceptual engineering, but none of that means we have discovered new and linguistically driven ways to procreate or become a parent. (140; emphasis mine)
Cappelen is thus insistent that we understand this object-level description of conceptual engineering as less philosophically provocative than it might at first sound. There are, to use his way of putting it, “ways of describing” the results of conceptual engineering that are both (a) object-level and (b) true. But the way in which they’re true doesn’t commit us to anything particularly philosophically weird, despite first appearances.
And yet, despite the philosophically innocent way in which we’re instructed to interpret the claim of conceptual engineering’s worldliness, it’s this object-level way of construing conceptual engineering that Cappelen adopts as part of his overall theory of conceptual engineering. Cappelen spends a full chapter arguing for the worldliness of conceptual engineering. Near its conclusion, he considers a Williamson-style skepticism about philosophical attention to concepts and conceptual analysis. He says:
Someone attracted to this [Williamsonian, insistently first-order] picture of philosophy would and should be prejudiced against the very idea of conceptual engineering. On the face of it—in the very name!—it seems to be advocating concept-fiddling, which is just what should be avoided. My version of conceptual engineering, however, isn’t like this at all: it is directly about the world, there are no concepts to fiddle with, and there isn’t any fiddling. […] In that sense, the process I describe as conceptual engineering is about knowledge, freedom, what is right, women, marriage, and salad. (146–47)
So a certain kind of philosophical needle-threading is being attempted here. That the engineer who advocates a change in the meaning of the word ‘salad’ is, ultimately, engineering salad itself is not meant to suggest anything particularly philosophically surprising or strange. And yet the description of them as engaged in this first-order, semantically unascended project is the official line, the response to the Williamsonian critic of conceptual analysis (see Williamson Reference Williamson2007), the most appropriate way to describe the project even if other ways—like the linguistically focused option (2) above—are, technically, also correct.
Of course, selecting the best or most philosophically illuminating way to describe something from a set of options that are all technically accurate is conceptual engineering par excellence! So it would be a mistake to write off Cappelen’s preference for option (3) over option (2) as peripheral to his overall view. But the tension between the presumably philosophically meaningful advocacy for option (3), on the one hand, and the disavowing of what seem to be its most striking implications, on the other, is symptomatic of what I’ll argue is a more general issue with Cappelen’s response to the Strawsonian objections.
3. Two different objections
Cappelen carves up the Strawsonian challenge into a number of different objections. But in his categorizing, he doesn’t distinguish versions of the objection that concern our ordinary judgments from versions that concern philosophical theses stated in a theoretical register. Cappelen thus fails to consider an important and distinct version of the worry. So contrast the following two worries:
(1) Error Theory: If conceptual engineering takes place, then the representational tools we use to make certain claims will have changed their intensions and extensions. But, if that happens, then many of our ordinary judgments about continuity of inquiry and substantiveness of disagreement will be false.Footnote 9 And yet, in many of the cases in which conceptual engineering is said to have taken place, it seems entirely correct to say that speakers before and after the purported change “said the same thing,” “talked about the same thing,” “really disagree with one another,” etc. The view that conceptual engineering has taken place would thus commit us to an error theory about many of our ordinary stability judgments and claims. Error theories are a cost; our stability judgments in these cases seem quite unproblematic, and so theories should avoid repudiating them if possible.Footnote 10
(2) Instability of Content: If conceptual engineering takes place, then the representational tools we use to make certain claims will have changed their intensions and extensions. But if those representational devices change their intensions, then speakers before and after the change will semantically express propositions that are about different kinds of things. Similarly, speakers who express disagreement using the respective meanings will semantically express propositions that are logically consistent. And yet, in many of the cases in which conceptual engineering is said to have taken place, inquiry is continuous, and disagreement is substantive. If the intensions of the relevant terms had remained constant in these cases, we would have an account of what makes the inquiry continuous and what makes the disagreements substantive—the speakers inquire into the same kinds of things; they express propositions that are logically inconsistent. But if the conceptual engineer posits a change in intension in these cases, then there is no way for her to explain what it is that’s continuous about continuous inquiry, and what constitutes the substance of a substantive disagreement.
These two objections are not meant to exhaust the possible “changing the subject” worries that someone could raise. Nor are they meant to be stated in such a way that anyone sympathetic to them would sign on for the exact versions stated here. Rather, they’re intended impressionistically, to make salient two different spirits in which the various worries can be felt and stated. On the one hand, they could concern a seeming conflict between conceptual engineering and our ordinary judgments as we express them in ordinary language. On the other hand, they could concern a seeming conflict between conceptual engineering and our considered philosophical views stated in a theoretical register.
Cappelen’s samesaying and coarseness of topic arguments are, I submit, well positioned to reply to the objection in (1). The Error Theory Objection has to do with the question of whether certain ordinary judgments and ordinary language claims can be vindicated. Cappelen’s arguments are, correspondingly, built up from claims about the semantics of ordinary English—chiefly, the observation that when we disquotationally report the speech of users of context-sensitive terminology, these reports do not seem to require identity of intension. If you’re worried about whether our ordinary stability claims come out true, then it should matter a lot to you exactly what we’re saying when we make those claims. Cappelen shows that what we’re saying when we make those claims is not something that requires identity of intension. Therefore, those claims do not go false just because the intension has changed. So long as sentences like “Speakers A and B said the same thing/inquire into the same thing/really disagree with one another,” etc. have true readings in English even when the intensions of the relevant terms have changed, then the conceptual engineer is not committed to an error theory about those claims simply in virtue of positing a change in intension.Footnote 11
The objection in (2) is different however. The Instability Objection is not about our ordinary judgments or intuitions. It’s not about whether there is some sense or other of “topic” in which Speakers A and B could be said to be talking about the same topic. It’s just directly a worry about intensions—a technical notion denoted by a term of art. It’s the worry that if speakers engaged in some inquiry refer to different kinds of objects (because of a change in intension), then they’re not investigating the same thing. It’s the worry that if two speakers express logically consistent propositions (because of variation in intension), then it’s not clear in what sense they express a disagreement. However, we might describe these cases in ordinary English, the conceptual engineer has, according to this objection, failed to give a theoretical account of how the relevant inquiries could be considered “continuous” and the relevant disagreements could be considered “substantive.” Saying that these cases are such that, in certain contexts, we can get away with using the ordinary English phrase “the same topic” is simply not responsive to this worry.
Consider for example a debate about salad. Our conceptual engineer claims the meaning of ‘salad’ has changed over time. The conceptual engineer’s opponent claims the meaning of ‘salad’ has remained constant over time. Suppose that the conceptual engineer and the opponent agree, pretheoretically, that inquiry into the nature of salad is continuous and that disagreements about salad are substantive. When asked in virtue of what that continuity exists, and in virtue of what those disagreements are substantive, the conceptual engineer’s opponent has a clear answer: early and later speakers were speaking about the same things—the things that, relative to a circumstance of evaluation, are in the extension of the word ‘salad.’ If inquiry is productive, then speakers gradually make fewer false claims about those objects and more true claims about them. Similarly, theorists who disagree about what salad is really like disagree with each other because they disagree about the nature of those objects.
What can Cappelen say on behalf of the conceptual engineer to these questions? He might, as he does several times in the book, reply in a defiantly first-order way: The inquiry is continuous because they’re both investigating the nature of salad! Their disagreement is substantive because they disagree about what salads are like! But even Cappelen allows that what these responses semantically express is false. If the intension of ‘salad’ has changed, then there is no kind of object into the nature of which both parties to the inquiry inquire, or about which the various parties to a dispute make conflicting claims.
What about Cappelen’s claim that while these responses semantically express false claims, they also express (just not semantically) true claims—claims that are true in virtue of the ordinary English, coarse-grained notion of “topic” at work in our our samesaying and disquotational speech reports? Unfortunately, this doesn’t answer the question either. The worry (the disagreement worry, for example) was that if the words have changed in intension, then the propositions that parties to the debate express are, in the typical case, logically consistent. If they’re consistent, then what is the basis of the conflict between them? This is a worry about a philosophical view, stated using the technical terminology of philosophy of language and semantics. Learning that in everyday English, in certain contexts, speakers can truly say “they really disagree with each other” even though the intension has changed, will be cold comfort to the person with this worry. Why can they say that? What is the thing, if not the sets of objects they’re talking about, that’s continuous through inquiry and the substance of the disagreement? Cappelen rejects the idea that intensions are the theoretical tool that can play this explanatory role. But he does so without providing anything to replace it.
It’s important here to emphasize that the objection in (2) is premised on the idea that the only thing that could provide continuity to some inquiry or stability to some disagreement is identity of intension. That’s not something you have to agree with. It’s not something Cappelen agrees with. It’s not something I agree with. But if identity of intension is not playing that role, then what is? The point is not that this question can’t be answered. It’s rather that the question is coherent and that, arising as it does in the context of theoretical inquiry, it cannot be answered with reference to the truth conditions of ordinary English sentences containing homophonous terminology. The fact that Cappelen doesn’t provide an answer doesn’t make his view wrong. But it reveals that his account is incomplete.
This point also helps to illuminate the fact that the Error Theory Objection and the Instability Objection are not, and are not intended to be, fully independent of each other. (And, thus, that being worried about the Instability Objection does not require signing on for some implausibly clear dividing line between ordinary and theoretical language.) The two objections are related in the sense that if our ordinary judgments of samesaying, continuity, and disagreement are tracking something other than identity of intension, it would be nice to have an account of what it is they’re tracking. With his Contestation Theory, Cappelen explicitly declines to provide such an account. But why should we accept, ahead of time, that there is nothing interesting to be said here? If there were a response to the second objection—a theoretical account of what is stable in inquiries that are continuous and in disagreements that are substantive—it could help, or be helped by, an account of what we track with our ordinary stability judgments. As is often the case, what we care about in ordinary contexts might very well be, or be closely related to, categories that we decide on reflection are significant in the philosophy classroom, even as we recognize that the two are in principle distinct, and that analyses of them bear motivation by different sets of considerations. Whatever the connection between the two, there is a coherent, philosophical question to be asked here: What is the stable ground of our debates and inquiry if not identity of intension? Cappelen doesn’t answer that question with his samesaying arguments. But that doesn’t mean that no answer is available.
To see what such an answer might look like, consider a view that Cappelen rejects later in the book—the view that meanings can play certain functional roles in our practices, and that the question of which meaning is better suited to playing a particular role is the kind of thing that speakers might inquire into or disagree about.Footnote 12 It’s beyond the scope of this paper to stump for this view, or to defend it from Cappelen’s criticisms. What I want to observe is simply how such a view is the kind of thing that in principle can provide an answer to the Instability Objection. So consider debates about the nature of marriage. The conceptual engineer might posit that part of what’s at issue in these debates is what the intension of the word ‘marriage’ should be. The conceptual engineer might even suggest that as a result of these debates, the intension of ‘marriage’ has changed over time. Despite this change in intension, the debates—or some of them, at least—strike us as forming a coherent, continuous conversation over time. If the inquiry is continuous, then what is the thing being talked about? It’s obviously not the objects or relations in the set that is the extension of the word ‘marriage,’ relative to a circumstance, as it is used by all the various parties to the debate. Because intensions differ from speaker to speaker, there is no such set. So what are they all talking about? Cappelen would say, “What they’re talking about is marriage!” If you find that answer unsatisfying, then you share the worry I’ve tried to elucidate here.
What the functional role allows us to do is specify a thing that—when the debate is sensible—all parties to the conversation care about, and that does remain a stable subject throughout, even as the intension of the word ‘marriage’ changes over time. Why argue about “marriage” at all? Well, the things we categorize together with that term play an important role in our lives and social practices, and so it matters which things we’re willing to lump together in that category. The things we agree to call a ‘marriage’ are treated as deserving of special kinds of esteem or deference, they have a special legal status, they play a meaningful role in the lives and life plans of many people. Against this background of discourse and social practice—that is, given the functional role this category plays for people like us—it matters a lot how we choose to use the word (in whatever language we happen to be speaking) that happens to designate it. Different choices represent different intensions, different determinations of which kinds of things should be playing the “marriage role” in our lives. So it makes sense that parties to the debate about the nature of marriage would care very much about how the word ‘marriage’ gets used, even if different speakers, in fact, refer to different sorts of things with it.
What makes the inquiry continuous—when it is—is that throughout the debate speakers hold shared assumptions about the role that the category of “marriage” (the things we agree to call ‘marriages’) plays for us. What they’re trying to determine, and what they disagree about, is which kind of thing is best suited to play that role. When those shared assumptions are in place, inquiry will strike us as continuous, and disagreements will be substantive. Of course, those shared assumptions are not always present. But this predicts, correctly, that to the extent speakers hold different assumptions about the very point of having a category “marriage,” their inquiry will strike us discontinuous, and their disagreements will strike us lacking substance.
This might not be the right view. Even if it’s on the right track, much is left to be filled in, and it could be elaborated in different ways.Footnote 13 But what this sketch of a view demonstrates is what it would look like to have a response to the Instability Objection. To respond to that objection, we must ask, “what does the continuity of continuous inquiry consist in, and what does the substance of substantive disagreements consist in?” Call this the “Stability Question.” That our words mean the same thing is one answer to the stability question—one theory of what forms the stable ground of our conversations and debates. It happens to be among the most widely accepted answers at the moment.Footnote 14 The conceptual engineer rejects that theory. Responding to the Instability Objection requires a theory of what replaces it.
Simply observing that, in certain contexts, the sentence “the speakers are talking about the same thing” has a true reading in English doesn’t provide such a response. The demand was not for the vindication of ordinary judgments or claims, but rather for a substantive answer to the stability question—for a theory. By contrast, the claim that the various speakers share a set of background assumptions about the functional role a certain category plays for us, and they seek to determine which kind of thing—which intension—is best suited to play that role is an answer to the stability question. It offers up a substantive description of a (fine-grained) topic interest in which is stable throughout inquiry and across disagreements, even as the intensions of the relevant terminology change. It—the functional role about which the speakers share background assumptions—might even be the thing that’s providing a true reading to sentences like “the speakers are talking about the same thing.”
4. Engineering “topics”
The issue revealed here isn’t necessarily a flaw in Cappelen’s framework. But it is, at the very least, a gap. Cappelen’s arguments have traction against one sort of Strawsonian worry but not against another. He thereby leaves unexplored a range of intriguing and important philosophical issues faced by the conceptual engineer. This theoretical gap exists because Cappelen treats the truth of pretheoretic claims like “the speakers are talking about the same topic” as a response to the philosophical worry that “the speakers are not talking about the same topic.” But, however tight the connection between our ordinary beliefs and our considered philosophical views, these two claims are not meeting head on.
If this situation sounds a bit like a recapitulation of Strawson’s original worry—like offering a physiology textbook to someone with a broken heart—it’s because it is … only backwards. Strawson’s analogy has the misguided conceptual engineer offering a pointlessly scientific answer to a question posed in the homophonous language of everyday life. In Cappelen’s case, it’s the other way around. It’s the question that’s posed in the language of theoretical inquiry, and Cappelen’s answer that is motivated by and framed in the language of everyday discourse. Cappelen, doubling down on his own reply, could, of course, respond that although the coarse-grained notion of “topic” at work in his answer and the fine-grained notion of “topic” at work in the question have different intensions, the question and the answer are nevertheless about the same thing. Topics! But that doesn’t feel like much of a reply. It could turn out that to ask the question in fine-grained terms is nonsense or undeserving of an answer. But the question seems sensible, and it seems deserving of an answer framed in the same terms as the question. The functional-role view demonstrates that the potential for such an answer is there.
The issue with Cappelen’s response to the Strawsonian challenge carries through to his positive account, and diagnoses what was unsatisfying in the claims of the “worldliness” of conceptual engineering. Worldliness sounds interesting as a philosophical thesis because we interpret it as coming out of a particular kind of philosophical context—a context where topics are finely individuated and, thus, where, if we “change the nature of F’s,” it can be inferred that there are some F’s in the world that we’ve thereby effected a change in. Cappelen encourages this kind of reading when he employs the worldliness thesis in responding to Williamsonian skepticism about a focus on concepts as opposed to the world itself.
And yet, as Cappelen is careful to point out, the reading that makes the worldliness thesis true is not a reading with these implications. ‘We’ve changed the nature of families’ is true only on a reading that’s “loose” in at least two senses: The true reading is not the proposition semantically expressed by an utterance of that sentence. And the true reading involves the looser, coarse-grained notion of topic. What semantically expresses a true proposition is just this: ‘we’ve changed the intension of the word “family.”’
This should strike us as a red flag. What seem to be the most interesting philosophical implications of the thesis do not follow when that thesis is interpreted in such a way as to be correct. In other words (in the philosophical context, at least) the claim is either interesting or true, but not both. How is the Williamsonian critic supposed to feel when, back in the philosophy classroom, we allow that, technically, it’s not actually true that what families are has changed, and that, technically, we’ve just changed the meaning of a word? I don’t agree with the Williamsonian critique of conceptual engineering to which Cappelen is responding. But I wouldn’t blame the Williamson critic if in this case they started to wonder if it was all just “concept fiddling” after all.
The question comes down to the uses to which we put the various notions of “topic.” Cappelen argues persuasively that the coarse-grained notion of topic is at work in our disquotational speech reporting and our ordinary stability judgments and claims. This observation has important philosophical consequences. I believe it serves as a genuine and successful reply to the ordinary language version of the Strawsonian challenge. But on discovering this coarse-grained notion, Cappelen, qua conceptual engineer, should immediately proceed to ask: “Is this the notion of topic we should be using?” When it comes to those ordinary speech, samesaying, and stability reports, the answer is unquestionably yes. A coarse-grained notion of topic is essential to those practices; they would be untenable if saddled with more fine-grained standards of correctness. But the coarse-grained notion itself is not a good fit with every context and, in particular, not with a philosophical context where we speak and theorize in intentionally fine-grained terms. Cappelen’s own arguments demand we seek to improve on it.
Acknowledgments
Many thanks for helpful discussion to Herman Cappelen, Andy Egan, Megs Gendreau, Justin Khoo, Dan López de Sa, Tristram McPherson, Sebastiaan Meissner, David Plunkett, Kristian Sheeley, Amie Thomasson, and Lila Wakeman.
Timothy Sundell is an associate professor of philosophy at the University of Kentucky.
