What is good thinking? Comments on Mona Simion’s Shifty Speech and Independent Thought

ABSTRACT

Mona Simion's Shifty Speech and Independent Thought argues for epistemic independence—the independence of good thinking from practical considerations. Along the way, she argues against “shifty” views of knowledge and knowledge ascriptions, as well as against those who have tried to preserve the independence of knowledge from practical considerations by accepting the dependence of the epistemic normativity of assertion on practical considerations. In my discussion, I start by highlighting some of Simion's main claims and reconstructing her main lines of argument. I then raise some minor concerns about her underlying methodology and some bigger picture concerns about some of her background theoretical assumptions and her argument for epistemic independence.

KEYWORDS:

• Mona Simion
• epistemic norms
• knowledge
• assertion

Mona Simion’s Shifty Speech and Independent Thought is clear, engaging, and covers an impressive amount of ground in around 160 pages. Simion’s main aim is to argue for the ‘independence of good thinking from practical considerations’ (Simion 2021, xi).¹ Her book is, as she puts it, a ‘manifesto’ for the independence of the epistemic domain from practical considerations. Specifically, it defends ‘classical invariantism’ (the view that knowledge does not depend on practical considerations such as stakes) against the ‘shifty’ views of knowledge (e.g. pragmatic encroachment or ‘sensitive invariantism’) or knowledge ascriptions (e.g. contextualism) that have been widely discussed, if not so widely endorsed, in epistemology for the past thirty years or so. It also argues against those classical invariantists who have felt compelled to adopt a shifty view of the epistemic normativity of assertion and other constative speech acts.

Simion’s central focus is one prominent argument for shifty views that is based on the knowledge norm of assertion. The knowledge norm says, roughly, that one’s assertion (that p) is epistemically proper if and only if one knows (that p). The argument is that epistemically proper assertion depends on practical considerations (e.g. whether one can assert that p depends on how much is at stake) and so knowledge must itself depend on practical considerations. Simion seeks to undercut this argument by arguing that, despite initial appearances, epistemically proper assertion does not depend on practical considerations. More broadly, she argues that epistemically proper speech does not depend on practical considerations, so one cannot argue that epistemically proper thought depends on practical considerations on the grounds that epistemically proper speech so depends. Simion therefore undercuts a central argument for shifty views of knowledge and/or knowledge ascriptions and so a central argument against epistemic independence.

If her aims were merely negative, this book would be a valuable corrective to a tendency to move too quickly from intuitive judgements about what it would be proper to say to conclusions about what it is to have knowledge (or, for that matter, justified belief). But Simion’s aims aren’t merely negative. She also defends a picture of the normativity of assertoric and other ‘constative’ speech-acts such as conjecturing. On this picture, these norms are grounded in the function of these speech-acts. In the case of assertion, the function is to generate knowledge in hearers, and so assertion is governed by a knowledge norm. (The details may differ for other constatives, but knowledge is still centrally important). The result is an attractive picture of the epistemic normativity of speech that is designed to deliver a classical invariantist picture of knowledge, and the independence of thought from practical considerations. This picture provides further support for Simion’s argument against shifty views like contextualism and pragmatic encroachment, and against the idea that good thinking depends on practical considerations.

In my discussion I will do three things. I will start by highlighting some further claims Simion makes and fill in some of the details of my rough sketch of her argumentative strategy (§1). I then raise some relatively small concerns about some of details of Simion’s argument, most of which relate to her underlying methodology (§2). I finish by highlighting two bigger picture issues, one of which relates to her functionalist framework for epistemic normativity (§3), the other of which relates to her central claim that good thinking is independent of practical considerations (§4).

1. Simion’s main claims

Let me highlight some of Simion’s main claims:

1. If epistemically proper assertion depends on practical considerations, then we either need to endorse a shifty view of whichever epistemic status is required for epistemically proper assertion (e.g. pragmatic encroachment about knowledge) or we need to endorse a shifty view of the norms of all constative speech-acts. (Simion calls this the ‘shiftiness dilemma’).

2. The shiftiness dilemma can be avoided because epistemically proper assertion does not depend on practical considerations.

3. If we view epistemic normativity through a functionalist lens, there is no problem combining classical invariantism with a (non-shifty) knowledge norm for constative speech acts.

Broadly speaking, I agree with all of these claims. I will briefly comment on each in turn. First, Simion is hardly the first to note that there is a seeming tension between classical invariantism and the knowledge norm of assertion (see DeRose 2002). But her aim in Chapter 1 is to show that the point generalises for any epistemic status governing assertion and to all constative speech-acts. The upshot is that the classical invariantist cannot afford to ignore the dilemma. In Chapters 2, 3, and 4 she examines some ways in which classical invariantists have tried to respond to the dilemma and finds them all wanting. While she raises some important challenges for these alternatives, I think she dismisses some of them a little too quickly (see §2).

Second, Simion thinks that the shiftiness dilemma can be avoided because it trades on a confusion between epistemically proper assertion and assertion that is proper all things considered (see Chapter 5). For example, in a high stakes bank case it may well be all things considered improper to assert that (you know that) the bank is open. But that doesn’t mean it would be epistemically improper to do so and it is this further claim that you need to get the shiftiness dilemma off the ground.

More generally, Simion thinks the shiftiness dilemma trades on a confusion between epistemic norms and norms with epistemic content. The knowledge norm of assertion is an epistemic norm – it specifies the epistemic position one must be in with respect to a proposition for one to be able to assert it with epistemic propriety. But non-epistemic norms can have epistemic content. For example, it might be a norm of firing canons that one should fire one’s cannon if and only if one knows the enemy is in range. But, while it includes the word ‘knows’, this isn’t an epistemic norm, and any intuitions we might have about when firing cannons is appropriate don’t tell us anything about the nature of knowledge or the semantics of knowledge ascriptions (cf. Hazlett, McKenna, and Pollock 2012). If it turns out that proper cannon firing is sensitive to practical considerations this hardly shows that knowledge is sensitive to practical considerations. The question is why we should think that the fact that our intuitions about the appropriateness of assertion are sensitive to practical considerations shows that knowledge depends on practical considerations.

While Simion is clearly right that there is a difference between all things considered propriety and epistemic propriety, it is important to note that this does not by itself undercut the shiftiness dilemma or undermine the arguments of those who use it to motivate rejecting classical invariantism. One might respond to Simion by arguing that our intuitions about the propriety of assertions do distinguish between epistemic and all things considered propriety (see §2).

This is where her third claim comes in. In Chapter 6 Simion develops a functionalist picture of the epistemic normativity of assertion. Roughly: The etiological function of assertion is to generate knowledge in hearers and assertions normally generate knowledge in hearers by virtue of assertors having knowledge. Epistemically proper assertions are assertions that generate knowledge in the normal way and so knowledge is the epistemic norm of assertion. Of course, assertions serve other (moral, prudential) functions, which generate other (moral, prudential) norms of assertion. Those norms may be sensitive to practical considerations and sometimes their associated functions may be more important than generating knowledge. In such cases assertions that are epistemically proper may well be all things considered improper. While I agree with Simion that the functionalist framework for epistemic normativity is elegant, I’m not so sure that the knowledge norm falls out of it (see §3).

In Chapters 7, 8, 9, and 10 Simion extends her functionalist account of assertion to other constative speech acts (conjectures, tellings, predictions, advisements, appraisals, etc.). She singles our conjectures, tellings, and moral assertions as requiring special treatment. Conjecture requires special treatment because many think it is very different from assertion, at least from an epistemic perspective, because it is far less epistemically demanding. Simion argues, against this, that in many ways conjecture is more epistemically demanding than assertion. Proper assertion (that p) requires knowledge (that p) whereas proper conjecture (that p) requires that one knows that one has warrant, though not sufficient warrant, to believe that p. Conjecture is therefore a far more intellectually demanding speech act than assertion. For example, young children can properly assert, but it is unclear that they can properly conjecture.

Tellings require special treatment because some think they are importantly different from assertions, and so governed by quite different norms. Simion disagrees. Even if tellings differ from assertions in that they are always directed towards an audience whereas assertions need not be, most assertions are in fact directed towards an audience, and on Simion’s functionalist framework this means they will be governed by the very same norm.

Finally, moral assertion requires special treatment because, while assertions typically have the function of generating knowledge in hearers, Simion suggests that moral assertions have the function of reliably generating moral understanding. As a result, the norm of moral assertion differs from the norm of assertion: a moral assertion (that p) needs to be knowledgeable and accompanied by a contextually appropriate explanation (why p). I think what Simion says about moral assertion is plausible, but it is indicative of a problem with her defence of the independence of good thinking from practical considerations (see §4).

2. Methodology in epistemology

In this section I will offer some brief comments on Simion’s discussion of the shiftiness dilemma. As we will see, most of these comments have to do with Simion’s underlying methodology. While I have some quibbles with what she has to say here, I don’t think any of the issues I raise in this section make that much trouble for her overall argument.

Let’s start with the shiftiness dilemma itself. As discussed in the previous section, Simion thinks those who use this dilemma to motivate rejecting classical invariantism or the knowledge norm of assertion confuse epistemic propriety with all things considered propriety. In doing so she appeals to the claim that our intuitions about cases, such as they are, don’t really distinguish between epistemic and all things considered propriety:

given the fact that actions, including assertion, are governed by many norms the requirements of which may conflict in a given case, how are we to distinguish the intuition triggered by the requirements of the norm we are interested in - that is, the distinctively epistemic norm governing assertion – from the intuition triggered by the requirements of other norms governing it, especially when the requirements are in conflict? (p. 66)

There is clearly something to this. In general, one shouldn’t expect our intuitions to be particularly sensitive to subtle philosophical or linguistic distinctions. That said, standard defences of the knowledge norm of assertion seem to assume that we are sensitive to the difference between epistemic and all things considered propriety. Take the argument (made by Williamson 2000 and repeated by many others) that we criticise assertions on the grounds that the speaker didn’t know (Why did you say the bank is open? You didn’t know!). Criticisms of this form may often be all things considered improper, or at least prudentially improper (they may not serve our purposes terribly well, as they can be quite annoying). If Williamson’s argument works, it works because we can set the fact that criticisms of this form are often all things considered improper to one side and recognise their epistemic propriety,

More generally, many arguments for the knowledge norm are based on intuitions about the propriety of assertions or criticisms of assertions. The knowledge norm is supposed to gain support from the seeming absurdity of asserting sentences of the form ‘p, but I don’t know that p’. It is supposed to gain further support from the fact that we can defend our assertions by citing the fact that we know the facts in question (‘Why did I say the bank is open? Because I know fine well that it is open!’). If these arguments work, they work because our intuitions about the (im)propriety of these and other assertions can somehow ‘zero in’ on epistemic propriety with some degree of accuracy. For example, there may be contexts where asserting sentences of the form ‘p, but I don’t know that p’ is all things considered proper (someone holds a gun to my head and tells me that, unless I assert ‘it’s raining, but I don’t know that it’s raining’ they will shoot me), or contexts where citing the fact that you know the facts in question is all things considered improper (perhaps your friends are fed up with you always trying to show that you were right). None of this tells against these arguments for the knowledge norm precisely because we can distinguish between epistemic and all things considered (im)propriety.

These points need not be fatal to Simion’s case against the shiftiness dilemma. But I take them to show that the case against it is a good deal messier than she supposes. Something similar applies to her discussion of various ways of trying to avoid the shiftiness dilemma in Chapters 2, 3, and 4. In these chapters Simion argues that epistemic WAMs (warranted assertability manoeuvres), pragmatic WAMs, and ‘KK compatibilism’ all fail to resolve the dilemma. While the details vary, her basic strategy is to identify some linguistic data that the view in question is unable to deal with. This is fair enough, but it belies the important point that there is room for debate about what exactly the ‘data’ are, and about which bits of data are most in need of explanation. Anyone familiar with debates about classical invariantism, pragmatic encroachment, contextualism, etc. will know that there is a huge amount of ‘data’ and there is no one view that is supported by all of it. Partisans of the various views argue about which view does the worst job of dealing with the ‘data’, and about which bits of ‘data’ really need explaining. While I’m not (now) a partisan of any of the views Simion discusses in these chapters, I suspect partisans will find plenty of places where they want to push back against Simion’s arguments.

3. Simion on epistemic normativity

The main reason why I don’t think any of the points raised in the previous section are fatal to Simion’s argument is that she has her own positive picture of epistemic normativity and the normativity of assertion. This picture is an alternative to the picture on which the shiftiness dilemma arises, and reasons to adopt Simion’s picture are further reasons to dismiss the dilemma. In this section I will take a more critical look at Simion’s picture of epistemic normativity, which she develops in Chapter 6.

To start, Simion’s claim is that the function of assertion is to generate testimonial knowledge in hearers. By ‘function’ she means etiological function or ‘e-function’. To say that assertion has the e-function of generating testimonial knowledge in hearers is not to say that assertions always generate testimonial knowledge in hearers. Rather, it is to say (roughly) that we make assertions because, often enough, they generate testimonial knowledge in hearers. Compare: to say that the heart has the e-function of pumping blood is not to say that all hearts pump blood (some hearts have stopped working). Note that Simion is talking about the epistemic e-function of assertion. Assertion may have many other functions and e-functions, but it (usually) has one epistemic e-function (generating testimonial knowledge in hearers).

I am a little puzzled why Simion’s view is that the (epistemic, etiological) function of assertion is to generate testimonial knowledge. Why not just knowledge? You can, after all, get knowledge (that p) from my assertion (that p) without that knowledge being testimonial knowledge. Imagine I assert that the wandering albatross has the largest wingspan of any non-extinct bird species. You think ‘huh, is that true?’ and go and check online. A quick google confirms that, while there may have once existed a bird species with a larger wingspan, the wandering albatross does indeed have the largest wingspan of any non-extinct bird species. In this example, you know that the wandering albatross has the largest wingspan of any non-extinct bird species because I told you (if I hadn’t told you, you would never have formed a belief about which bird has the largest wingspan) but you don’t simply believe this in virtue of me having told you. This doesn’t seem like an unusual sort of case. While assertions often generate testimonial knowledge in hearers, they also sometimes prompt hearers to go and do their own investigations. Why focus on the former sort of case and ignore the latter sort?

Perhaps Simion is assuming that, while we sometimes verify testimony for ourselves, we usually don’t. Normally, assertions generate testimonial knowledge, rather than prompting hearers to go out and independently verify the truth of what has been asserted. Note that, in saying this, Simion would not necessarily be taking a side in the reductionism/anti-reductionism debate in the epistemology of testimony. Assertions may generate testimonial knowledge in hearers because hearers monitor speakers for trustworthiness, in the spirit of Fricker (1994). The point is just that, normally, assertions generate testimonial knowledge rather than prompting hearers to go out and gain that knowledge for themselves via observation, reasoning, or what have you.

This might explain why Simion needs to make special accommodations for moral assertions in Chapter 10. Many think that there is something wrong with simply forming moral beliefs on the basis of moral assertions. Following Alison Hills (2009), imagine Eleanor forms the belief that eating meat is wrong because her normally trustworthy friend tells her that eating meat is wrong. Even if we assume it is true that eating meat is wrong, and that Eleanor’s friend is reliable about moral matters, there is – some think – something wrong with how Eleanor has formed her belief. One tempting diagnosis of what has gone wrong with Eleanor is that she may know that eating meat is wrong, but she doesn’t know why. (If you think eating meat is not wrong, or even morally right, simply change the example accordingly).

While she doesn’t necessarily endorse this diagnosis, Simion is happy to work with it. As outlined in §1, she proposes that the function of moral assertion is to generate moral understanding and so she ends up with the following norm of moral assertion:

At a context C, one’s moral assertion that p is epistemically permissible only if (1) one knows that p and (2) one’s assertion is accompanied by a C-appropriate explanation why p (p. 154)

Some quick comments on this. First, Simion specifies that proper moral assertions are accompanied by contextually appropriate explanations. The thought is that the form of explanation required depends on aspects of the situation or context in which the explanation is proffered. This is plausible because, in general, (good) explanation is context-sensitive in this way. Whether an explanation is good or not depends in part on what we need it to do – on our interests, purposes, etc. As I go on to discuss in the next section, this might cause trouble for Simion’s central aim, which is to argue for the independence of good thinking from practical considerations.

Second, the reason why Simion thinks moral assertions call for special treatment is that there seems to be something wrong with simple deference in the moral domain. You shouldn’t form moral beliefs by simply consulting ‘moral experts’. If this is right, though, it seems to me that a similar point will apply to other kinds of assertions. Take, for example, scientific assertions. While it may be perfectly fine for laypersons to form scientific beliefs based on simple deference to scientific experts, this is not so when it comes to scientists themselves, or to testimony within the scientific community. Or take philosophical assertions. Should you believe my criticisms of Simion’s book based on my simple say-so? (If you don’t think I’m a genuine expert, substitute the arguments of someone who you do view as an expert). And what about aesthetic assertions? Should you believe a painting is beautiful or a book seminal simply because a critic told you so?

In all these cases, it seems like something similar to Simion’s norm of moral assertion is required. For example, you might say that proper scientific assertion (at least, proper assertion within the scientific community) requires the provision of contextually appropriate explanations (cf. Gerken 2022). The more domains in which simple deference to experts isn’t appropriate, the more special norms of assertion we will need, and the less clear it becomes that the function of assertion is to generate testimonial knowledge.

Let me now turn to a related concern about Simion’s claim that the function of assertion is to generate testimonial knowledge and her use of this claim to defend the knowledge norm of assertion. To recap, she claims that knowledge is the norm of assertion because the normal way in which assertion generates knowledge is by being knowledgeable. But why is this the normal way in which assertion generates knowledge? Simion thinks it is the normal way because knowledge is readily available (see Chapter 6, §6.2). Knowledge is easy enough to get, and easy to transfer to others through asserting. Of course, there may be situations where an assertion generates knowledge in some other way, but these are the exceptions.

Whether this picture is true or not depends in part on the issue above viz. whether moral assertions are exceptional in the way Simion thinks they are. But it also depends on whether Simion is right in thinking that knowledge is readily available. Is it particularly easy to get, for example, moral, political, philosophical, natural, social scientific, social, interpersonal, introspective, or memory knowledge? It is at least not obvious that it is. In some cases, the reasons why it is difficult parallel the reasons why it doesn’t seem like you can get moral knowledge from testimony. In others, the reasons might have more to do with the complexity of the underlying subject matter (e.g. scientific knowledge), or with ways in which our social environment is structured that foster kinds of ignorance (e.g. about the social world). Importantly, these worries can be motivated even if you think that we can have true beliefs within these domains. There are grounds for thinking that lots of our beliefs in these domains fall short of knowledge, even if they are true (see, for example, Carter and McKenna 2020 on political knowledge and certain kinds of natural scientific knowledge, or Kornblith 2012 on certain kinds of introspective knowledge).

Simion might respond that what she really means is that getting, for example, simple kinds of perceptual knowledge is easy enough. But then it is unclear what supports her claim that the function of assertion is, in general, to generate testimonial knowledge, or that assertions normally generate knowledge by being knowledgeable. We should not assume, in the absence of argument, that an account of the function and normativity of assertion that is based on the function and normativity of a certain kind of assertion (e.g. assertions of simple matters of fact that are easily recognised by everyone) will extend to other kinds of assertion, and in particular we shouldn’t assume it will extend to assertions of complicated matters that are hotly disputed among those who know the most about them.

4. Epistemic independence

Simion’s central concern is to demonstrate the independence of good thinking from practical considerations. I am going to finish by suggesting that, even if she is right that shifty views of knowledge and knowledge ascriptions should be rejected, there’s a difference between rejecting shifty views and showing that good thinking is independent from practical considerations.

While she doesn’t go into that much detail, Simion suggests that her background picture is similar to the picture of normativity defended by Ernest Sosa (2007) (see p. 12 ff. 3). Let me briefly summarise Sosa’s picture. For Sosa, ‘critical domains’ are organised around central goods, which are valuable for their own sake relative to the domain, and structure normative evaluations within the domain. For example, the domain of archery is organised around the good of hitting the target, which is valuable for its own sake relative to the domain of archery and structures the normative evaluation of archery shots. An archery shot is good (for an archery shot) if it hits the target. The epistemic domain is organised around epistemic goods (for Sosa, true belief; for Simion, knowledge), which are valuable for their own sake relative to the epistemic domain, and structure normative evaluations of beliefs (and judgements). A belief is good (for a belief) if it is true (Sosa) or knowledge (Simion).

The crucial aspect of Sosa’s picture is that these critical domains are ‘insulated’. The goods that structure these critical domains are valuable for their own sake relative to the relevant domain, but they are not necessarily valuable as such or all things considered. Most importantly, practical considerations may be relevant to whether a good that structures a critical domain is valuable all things considered, but they are not relevant to whether it is valuable relative to the domain. Practical considerations may be relevant to whether hitting the target in archery is all things considered valuable, but they aren’t relevant to whether hitting the target is valuable relative to the domain of archery. Similarly, practical considerations may be relevant to whether true belief or knowledge is all things considered valuable, but they are not relevant to whether true belief or knowledge are valuable relative to the epistemic domain.

With this in hand, we can see why you might not need to defend a shifty view of knowledge or knowledge ascriptions to think that good thinking is dependent on practical considerations. Let me try two ways of making the point. First, you might think that the central organising good of the epistemic domain isn’t knowledge (or true belief) but rather significant knowledge (or true belief). More generally, your view might be that good thinking is good thinking about things that matter, and it is hard to see how to give an account of what matters that is independent of practical considerations. Simion might respond that the epistemic good is simply knowledge, though of course what we want all things considered is significant knowledge. Fair enough, but then it seems like she hasn’t demonstrated the independence of good thinking from practical considerations. Rather, she has demonstrated the independence of epistemically good thinking from practical considerations.

Second, let’s take another look at Simion’s norm of moral assertion. As I noted in the previous section, Simion holds that proper moral assertions are accompanied with contextually appropriate explanations, and what makes an explanation contextually appropriate will depend on our interests and purposes in the relevant context. But then practical considerations, broadly speaking, partly determine whether a moral assertion is epistemically proper, which means that proper moral assertion is not entirely independent of practical considerations.

Of course, this sort of dependence on practical considerations is a far cry from pragmatic (or moral) encroachment on knowledge, or from a contextualist semantics for knowledge ascriptions. But it is important for at least two reasons. The first is that, if I am right that Simion is going to have to countenance norms for several kinds of assertions that advert to contextually appropriate explanations (recall §3), this minimal dependence of the epistemic on practical considerations will end up being quite extensive. For example, proper scientific assertion may depend on practical considerations in this way. The second reason is that it points to a problem with Sosa’s picture of insulated critical domains, or at least with Simion’s version of it. Simion wants to say that practical considerations are irrelevant within critical domains (this is what makes them insulated) though are relevant to whether the central organising good of a critical domain is worth pursuing all things considered. But, just as it is extraordinarily difficult to fully insulate a draughty, poorly built British home, it is extraordinarily difficult to fully insulate a critical domain. To fully insulate a critical domain, we would need to be able to spell out all the norms and value claims within the domain without using any vocabulary that is interest-dependent in the way that talk of ‘contextually appropriate explanations’ (or ‘significant knowledge’) is interest-dependent. Perhaps this is possible for some domains, but I doubt it will be possible for the epistemic domain.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Unless otherwise stated all references are to Simion (2021).