Abstract
In this paper I undertake to resolve a main pragmatic puzzle triggered by Bank-type cases. After accepting ‘sanitized’ intuitions about Truth-Values, as reflected in x-phi experiments, the pragmatic puzzle about whether the husband is inconsistent remains, and if he isn’t (contrary to a first-blush impression), which intuitively is the case, how are we to explain it. The context in such cases is pragmatic, with awareness of high risks, and the treatment I propose is pragmatic as well, but not Gricean. I offer a new Pragmatics whose main tools are Steering Thrust (towards, e.g., action) and Posting. It focuses on what I call Action-Directed Pragmatics. An overall view of Pragmatics centers on the primary resort to Sayability rather than to Truth-Values, and to the Pragmatic phenomenon of co-present operative norm-types, which can conflict in some cases vis-à-vis particular actions and about what subjects can say. The very satisfactory explanation of the above puzzle that I offer is in such terms. In particular, attention is focused on the phenomenon of Negative-Polarity Pragmatic Functors (such as: I don’t know that p, or: I am not sure that p; or: It might be that not-p).
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
This paper covers the central Pragmatic phenomenon associated with this type of puzzles. I therefore exclude cases of first-time encounters, which result in a major confidence loss. I also don’t deal here with shortcuts construals that fit this sort of puzzle. For more detail, and also for an account of how the Risk Bias affects this puzzle, see my (2015b) and (2015c).
Jason Stanley (2005).
Such as, for instance, Brian Weatherson’s (2005).
This, I take it, is a variation, though a pretty loyal one, on DeRose’s bank case, mutatis mutandis.
We assume here no cognitive failures on the husband’s part such as memory lapses. The very short time difference between the opposite responses without relevant information renders him inconsistent if he literally means what he explicitly says.
We can generate an inconsistency puzzle not just for the husband, but also for us—for how we recast the case, given the following intuitive responses of audiences to the case, including us (selectively), concurring with the husband’s responses:
At stage 1, we have concurred with: The husband knows that the bank is open on Saturday.
At stage 2, we assent to: The husband doesn’t know that the bank is open on Saturday.
However, the husband hasn’t received, between stages 1 and 2, any information evidentially relevant to whether the bank is open on Saturday.
(And we can add, although this is not necessary for the husband’s inconsistency—only for driving through that we have an inconsistency problem (or at least are defective in our second assent): We are aware of that (by now—explicitly, or even implicitly, or just gauge it).
Hence: We now come out as inconsistent: At stage 2 we negate, by our assent, what we have said at stage 1, and without the husband’s (nor us) having received any information relevant to the factual matter as to whether the bank is open on Saturday (nor relevant to the husband’s pertinent epistemic state regarding it).
But intuitively, neither the husband nor we seem inconsistent. On the contrary, his responses seem quite plausible, appropriate, and rational, as is our assent to his self-characterizations—in particular, about his saying that he doesn’t know (in stage II). This is a distinct puzzle from the original one, and its treatment is distinct, though closely related, and via the same Pragmatic approach I employ in this paper. I treat this version of the puzzle in my (2012d). (Note DeRose’s different contextualist treatment, which takes the husband and us as respective ascribers (or self-ascribers) in our two distinct contexts, and thus has some affinity to my treatment of this derivative puzzle.)
‘p’ is sayable by x (at t) in case x is in a position (then) to say that p—the analogue of ‘assertible’. I expand on 'being in a position' below -- in view of operative normative pressures.
See the end of footnote 9: In this paper I deal only with the assessment of the husband’s response, not with how we, or audiences in general, can respond, which I construe as a vicarious response. I treat this latter issue at length in my (2012d).
See Stanley (2005). See also Buckwalter (2010) and Turri (2017). Elsewhere I develop in greater detail what I consider to be the fallacy associated with a resort to locutions such as, e.g., ‘intuitively true’—which I called ‘the Quick-Jump Fallacy’; see my (2017). For repercussions of such fallacies for proper philosophical methodology, see my (2012a), Section 6, and my (2015a), Section 8.
In general, and particularly if they have opted (or have considered opting) for one action (of Delaying the Deposit) rather than the other.
That is, evidentially and rationally speaking—in terms of Epistemic Rationality. And, we assume, the husband hasn’t received, nor has he discovered, any information relevant to his epistemic or cognitive faculties that bore on his epistemic position on the matter at hand. And more specifically, nor has he received information that undermines his epistemic position regarding evidence for or against his epistemic position on the matter (i.e., whether the bank is open on Saturday) at either stage (e.g., where he would thereby be misled). I here treat ‘know’ (or Knowledge) traditionally, that is, as epistemic, i.e., as depending only on evidence (and Truth, and/or on Reliability or Indicativity or defeaters and the like, all of which are stakes-independent so long as you add (to the first two) supplements, e.g., ‘sufficient’, and paradigmatically (and not derivatively) in the case when it is used at ‘rest’ (i.e., in Low, or in retrospective positions), which is called for in view of my Pragmatic position that rejects Pragmatic-Encroachment. That is, there is a certain Sayability value for knowledge-ascriptions in ‘Low’, but it may vary under various Pragmatic pressures. ‘sufficiently’ (when appended to 'he doesn't know') may be taken as a suppressed modifier varying with Pragmatic pressures. This would be a suppressed-modifier conception rather than my favored treatment via a Sayability function, which is much more general, and on which 'sufficiently' here would be considered a suppressed Sayability modifier (i.e., not being in a sufficient epistemic position to say) rather than a knowledge modifier, strictly speaking. This, of course, is by contrast to the main move of Pragmatic-Encroachment protagonists of re-construing ‘know’ (or Knowledge) as depending on a pragmatic (‘practical’) component, which has been their main move in addressing the puzzle, by contrast to Pragmatic positions or moves (in the sense of Pragmatics), such as mine, that claim to handle such a puzzle without re-construing ‘know’ (or Knowledge). (I use here throughout ‘Pragmatic’, with a capital ‘P’, as an adjective of ‘Pragmatics’, which is a branch of Linguistics or Philosophy of Language—as distinct from ‘practical’, or from being construed as a derivative of Pragmatism.). For a different pragmatic treatment, see Gerken (2017).
Here I specifically have in mind the explicit content of what he said he knew, viz., the explicit content of the content clause.
As here, in switching from ‘I know that p’ to its literal negation—‘I don’t know that p’. Again, I of course assume that the husband, in between the two stages, hasn’t received any other factual information relevant to the case, and specifically to whether or not the bank is open on Saturday (and regarding his pertinent cognitive and epistemic functioning); see also the previous two notes. (By ‘factual’ I have in mind empirical facts.).
And then there was the other version of the puzzle regarding our inconsistency, brought out in footnote 9 above, which I do not have the space to engage with here. But see also below, last section.
I thus, at least terminologically, take ‘Pragmatic Encroachment’ to cover the views of, e.g., DeRose, Stanley, Weatherson, and Fantl & McGrath—for convenience, in this paper, even if it would seem to some not to be an accurate umbrella-cover (regarding one or more of the above). A more accurate term would be: broad Pragmatic Encroachment. Another candidate is: Anti-Purism. (But I avoid using Stanley’s nomenclature since it’s tilted pejoratively, and I doubt whether this is called for in such a scholarly debate.).
The puzzle may have been, on some versions more than on others, less crisp than its conceptualization here.
Here I focus on a variant of the case, along lines akin to DeRose’s original formulation, with two distinct stages and with the two speech-acts, in each stage, by the husband. Some have presented another variant of the case as divided into two distinct sub-cases (corresponding to the two stages, with two distinct, unrelated, speakers). The problem of inconsistency, or at least a serious epistemic violation, arises then for the two speakers, (as well as for us, the readers, with the same assumptions, mutatis mutandis, in a form of at least a logical flaw or incongruence—of two speakers with exactly the same pertinent factual information saying seemingly contradictory statements that seem true). I focus on a two-stage format since this was DeRose’s original formulation, and especially since in such a presentation the inconsistency riddle is particularly poignant. Note that in such a presentation there is an extra exacerbating element which is the husband’s awareness of his saying later what seems to be the opposite of what he has asserted before. DeRose subsequently withdrew his original format in favor of a much more complicated scenario in his (2009), which in my view mixes Pragmatic as well as epistemic issues. The format employed here is one that, I claim, presents a purely Pragmatic puzzle—neither epistemic nor semantic. In order to eschew any epistemic interference and yet preserve the core of the Pragmatic puzzle I omitted the husband’s evidence in the original puzzle (viz., that he has been in the bank recently twice on Saturday). Such an addition is not necessary for generating the puzzle.
I take such recapitulation to be coarse enough to fit the point of this paper, as well as cover, to a more-or-less sufficient extent, such diverse strands, even though it's not tailor-made for each strand, and even if it would be accordingly, to some extent or another, faulted by proponents of such particular strands who would require a better fit.
And there were of course other responses as well as attempts to fix problems of the approaches mentioned here—see, e.g., Bauman (2011, 2012), Blome-Tillmann (2020, forthcoming), Bach (2008), Brown (2005, 2006), Dinges (2016), Engel (2009), Kim (2016), Pritchard (2005), Reed (2010, 2013), Roeber (2020), and Schroeder (2012).
This has been the more common formulation—‘objective’, as distinct from ‘perspectival’ which is, I hold, the right position (though more cumbersome), on which different representations of stakes screen off the actual stakes vis-a-vis the subject's decision, action, and verbal response (if there are objective stakes; whether there always are such, and if so, what they are, has been controversial; contrast with Stanley’s ‘objective’ ignorant high-stakes). On my view, regardless of whether and when there are ‘real’ stakes, or, when there are such, whether they are ‘correctly’ represented (by the subject, or by us), the phenomena in question are assessed from a perspectival position: What needs to be explained are responses of actors/speakers, as well as intuitions about actors' responses and about audience responses, in view of news about stakes and how they might have changed. But the pertinent effects of news about stakes (such as responses and intuitions, of actors, speakers and readers) would be the same regardless of whether or not such news is correct: Such news, in such cases, as standardly presented (shared by the conversational participants and the readers), screens off actual stakes (assuming it is accepted by the conversational participants and the audience). (This is also one way in which I try to side-step here the issue of whether some such stakes are ‘real’.) The presentation of the riddle is simpler if we assume, as is standardly the case, that the stakes have changed or that they are thus-and-so. But a more accurate and more general formulation of the riddle regarding such responses by competent language speakers needs to be in terms of represented stakes (or represented new stakes). Such a formulation side-steps the issue of whether the stakes under consideration are ‘real’, or are accurately represented.
I allow myself to be lax in tending to the particular variations in the formulation of such different treatments, for ease of presentation.
DeRose later (see his (2009)) reformulated his example in a quite different and much more elaborate variation, thereby obscuring some of the Pragmatic elements. In my (2015a) I distinguish sharply between purely epistemic retractions, purely Pragmatic retractions, and mixtures thereof. In this paper I deal only with Pragmatic retraction, which is quite close to the type of retraction of the original bank example. The variation I provide here does away with dispensable error-possibilities (e.g., whether the evidence from the husband’s visits to the bank on Saturdays has been misleading vis-a-vis the opening hours), and as such it is a case of a purely Pragmatic retraction. (They are dispensable in that the crux of the puzzle remains intact without them.) DeRose, however, in that (2009) longer example, coats the example with further epistemic fog, which renders the retraction in that case significantly epistemic, and as such beyond the scope of the purely Pragmatic treatment provided here. Its analysis would require accounting for the Pragmatic aspects (as brought out here), the epistemic aspects, and how they interface. At the heart of my approach is the thesis that there are important Pragmatic aspects of retractions, which need to be accounted for via a suitable Pragmatics, and accordingly it would be methodologically better to consider them in relative isolation.
Or in terms of features of Knowledge.
See note 21 above.
And similarly with Epistemic Contextualism, with the requisite context dependence. ‘Shifty Epistemology’ is the term of Fantl & McGrath.
I have for many years advocated my anti-Pragmatic-Encroachment Pragmatic invariantist position to the effect that the threshold in the truth-conditions of Knowledge-Ascriptions is not contextual or stakes-dependent, but it is at least communal/dialectal: It features in the competence of the use of the term ‘know’ within particular linguistic communities and is stable in them (and may well be more universal than that—all of this is subject to empirical tests). However, my former particular pronouncement to the contrary in my (2006) “Probabilistic Knowledge” (p. 7, regarding the threshold of the KI condition) echoes interest-relativity. This pronouncement reflected and addressed what I took to be a widely acceptable sentiment at the time, but not a position that I carefully examined. It doesn’t reflect my present position (since around 2010). I thank here Rodrigo Borges for calling my attention.
I here lump together the above types of broad Pragmatic-Encroachment approaches, which of course does some injustice. However, here I attempt to focus on what is common to these approaches regarding how to deal with the riddle involved in bank-type cases and their kin.
Assuming, as I do, that he was candid in making these statements. Note, as will be clearer below, that the notion of inconsistency can be extended to corresponding defects in Epistemic Sayability (see the end of the paper) and more broadly, to incongruence, which covers inconsistency as a special case that is (between the semantic contents of two sentences (in context)), and also to cases where no Truth-Values are recognized or brought to bear (e.g., just implicit Action-Directed content, or even ethical contents, all of which I consider to be Steering Thrusts as well, hence Pragmatic inconsistencies—between Steering Thrusts; for the latter, see my (2020b)).
If a speaker hasn’t received broadly relevant information (i.e., in our case, relevant evidentially or to an impending action) between the time of saying one thing and then its negation (with the same contextual relevant respects—i.e., factual or chancy, but stake-wise sameness of context isn’t required—the context overall may be different), he may well be inconsistent (or incongruent, and thus logically faulty—if the conveyed contents are incompatible, e.g., epistemically inconsistent if the epistemic (conveyed) Steering Thrusts are incompatible, or AD-inconsistent if the AD Steering-Thrusts of his two sayings are incongruent with no received intermediate information relevant to the action in question): Such an incongruence doesn’t require that what is conveyed are truth-value carriers since it applies more generally to conveyed contents (explicit or implicit, but in any case, from my Pragmatic perspective, to Steering Thrusts—see the end of the paper). It also applies to signaling, or more precisely what I call Posting, whereby a speaker might commit himself to a Pragmatic stance, e.g., to an epistemic stance regarding some semantic content or a proposition, and thereby end up being epistemically and/or semantically inconsistent; but I won’t pursue this direction here.
Accordingly, in this vein, the notion of inconsistency, or better, incongruence, can be extended to other norms, e.g., ethical norms: If a speaker conveys the same ethical content and then its negation on the basis of the same available ethically relevant data, and if her possession of and access to such data hasn’t changed between her making the two judgments, then she is inconsistent or incongruent (or faulty). As will be clear from the discussion below, although I don’t have the space to elaborate on it in this paper, this (diachronic) inconsistency (or incongruence) holds when the relevant applicable normative pressures haven’t changed (between her two such speech-acts) and she hasn’t acquired in the meantime any pertinent information, in particular information made relevant vis-a-vis these operative and applicable normative pressures in the case in point.
So there can be in such cases logical inconsistency, or ethical incongruence, etc. But the remaining seeming incongruence under consideration here (once we have clarified that he is not semantically or epistemically faulty) is Pragmatic inconsistency, or incongruence, that is, due to conflicting Steering Thrusts (see the end of the paper).
So the mistake (in considering the husband inconsistent, in the sense of being committed to two incompatible contents) arises if one doesn’t take into account that different operative normative pressures are at play in the two instances which are relevant to the Sayability in question. So a semantic focus on (semantic) inconsistency in the puzzle (i.e., vis-à-vis the pertinent Truth-Values of the sentences he utters in their contexts) does not engage the only applicable notion of incongruence here since incongruence can be considered under different norm-types. The bottom-line of explaining away the seeming inconsistency, when construed as semantic inconsistency, as is normally the case, of the speaker (rather than the conveyed contents of the utterances in their contexts) is that only when the explicit (semantic) contents of the speech-acts in question and its negation are conveyed there are straightforward, formal grounds for real inconsistency (here, semantic/epistemic); but general incongruence is applicable to other norm-types as well. Yet the explicit content need not be conveyed when there are normative pressures that outweigh the semantic/epistemic pressure. (Contents that the subject is committed to should also be taken into account, but this concerns the topic of Posting, which is outside the scope of this paper; see my (2021a).) The charge of incongruence (or inconsistency) may therefore depend on whether there is a significant shift in relevant applicable normative pressures that transform the conveyed contents. (‘Relevant’ is sensitive in particular to normative changes regarding whether the sentence employed in its context conveys explicit or implicit contents or both, bearing in mind different types of implicit contents (here AD-implicit contents, but there can also be others, e.g., ethical).)
The Pragmatic approach offered below as applied to bank-type puzzles primarily addresses the question of how to construe the data on which the husband indeed seems to be ‘right’ in what he says (or that what he says seems right), and, more generally here, below and elsewhere, how to account for a speaker’s seeming ‘right’ in what she says with ‘intuitively true sayings/assertions’ (or ‘intuitive saying-truly’). There is a seeming, or prima facie, inconsistency in a case like ours, and in particular if what the speaker says expresses contradictory semantic contents. Yet the speaker is inconsistent, outright, if we appropriately construe the intuitions we have that what the speaker has said is right (in both sayings) semantically. Yet our intuitions about what a speaker says in contexts like ours target what the speaker conveys—whether explicitly or implicitly. And what he says explicitly need not cover what he conveys, and furthermore, what he says explicitly (on a given occasion) need not be conveyed. This is the heart of a Pragmatic approach in general—taking implicit conveyed contents into account, and checking (when it's called for) whether the explicit contents are conveyed—and to what extent. Ultimately, our intuitions about a speaker’s being inconsistent (or incongruent) in what she says are partly Pragmatic and target what she conveys. Thus, it might seem that a default and natural way of construing the speaker (as in the bank case) is just semantically—but of course it need not be, and in general wouldn't be, and the inference from ‘What the speaker says is right’ to ‘What the speaker says is true’ is in general fallacious, as it is in our case—the bank-type puzzle case. Such a construal. when covering implicit contents, requires not only that what is conveyed has a truth-value, but also that it is to be evaluated just semantically. However, what the speaker has said can be also evaluated Pragmatically, i.e., in terms of Pragmatics—in terms of the contents that the speaker conveys, which may well be implicit and not just explicit. When so construed, the conveyed content might not be a proposition, but a different type of content, such as a Pragmatic content—which here is Action-Directed content, in the form of Action-Directed Steering Thrust, which doesn’t have a truth-value.
This requires a different Pragmatics than Gricean-like Pragmatics—it requires in particular, in such a case, Action-Directed Pragmatics, on which conversationally conveyed contents can be implicit and not be propositions. The speaker then might be right (in saying what she does, or in what she says, which covers what she conveys) without what she says (in that context) being true, since what she says (in such Action-Directed contexts) is evaluated as steering toward the right action to take in the deliberation at hand. Accordingly, what is evaluated as ‘right’ in saying what she does (or in what she says) is her steering towards a certain action (and sometimes also how she does it—such as the degree of steering). Not keeping in mind or appreciating this distinction (primarily between two such distinct types of conveyed contents—propositional vs. Action-Directed contents, and more generally between the literal/semantic content and the Pragmatic/conveyed/implicit content) is a source of a common mistake of jumping from judging that the speaker is intuitively right in saying what she does (or in what she says) to concluding that therefore, what she says is intuitively true, or that therefore, intuitively, what she says is true and therefore is true. If what the speaker conveys is Pragmatic content which is not propositional, the inference is of course fallacious, since an Action-Directed message/content per se, which is not propositional, doesn’t have a Truth-Value and thus can’t be true nor can it be ‘intuitively true’. So even though often times it seems very plausible for us to move from evaluating what the speaker says as ‘intuitive’ to its being ‘intuitively true’, this inference is not in general valid, since it presupposes a semantic construal of what the speaker conveys: It fails when the content conveyed doesn’t have a truth-value, as in the case when it’s just Steering Thrust. But, of course, appreciating such a stance requires not taking Pragmatics, and specifically implicit contents, to be just Gricean-like, and in particular, to be propositional. The above perspective invites a different type of Pragmatics—Steering-Thrust Pragmatics. Therefore, what may seem not to allow for a Pragmatic construal, and thus requires a semantic construal, if taken from the perspective of a Gricean-like Pragmatics, need not be so from the perspective of Steering Thrust Pragmatics. Consequently, the fault I find in adopting Pragmatic Encroachment is not necessarily a fault in the reasoning of its proponents, given that their horizons were limited just to Gricean-like pragmatics. For greater detail about Steering-Thrust Pragmatics, see my (2018b), (2018c), (2018d), and (2020a). Elsewhere I develop the point about the above faulty transition further, and call it ‘the Quick Jump’ fallacy; see below; see also my (2017). For Grice, see, e.g., (1989).
I.e., features pertinent to the content clause: That the bank is open on Saturday; and accordingly, to the husband’s epistemic (and cognitive), in particular, evidential, position regarding it. Again, as noted above, I focus on the original two-stage version of the bank case, but the riddle and its treatment as elaborated here suit also a version with two distinct cases (different conversational participants, different times).
In the pertinent situations or contexts.
Again, such characterizations are rough and don’t do full justice to the variety of proponents of Pragmatic Encroachment, broadly conceived (e.g., Stanley’s conception of the metaphysics of Knowledge, or Epistemic Contextualism, with its suitable context-dependence, where counting-as-true is in an ascriber’s context.). I group them together under the title of ‘broad Pragmatic encroachment’. I settle on such a rough characterization in order to focus on what I take to be a common core position.
Whose value has changed.
Directly—non-contextually, as by SSI, or indirectly, via the ascriber’s context. A main distortion of the traditional view by Pragmatic Encroachment is in undermining the ‘stability’ of knowledge: This can be taken in the sense that the value of ‘know(Low)’ is expressed by uses of ‘know’ much less commonly, as is any other value of ‘know(i)’ (for the sub-senses of ‘know’), than ‘know’ as taken to be traditionally. But ‘know(Low)’ has largely similar applicability regarding target knowledge states from (sufficiently) retrospective perspectives, which are in Low, as is traditional ‘know’. From the Pragmatic perspective pursued here, what is less common is the Sayability of ‘know’ (specifically in the use of Skads -- Self-Knowledge-Ascriptions-Denials), which is indeed risk-dependent (i.e., we wouldn't often say that the subject knows even when the subject is in an epistemic position to tell that she knows, whereas in retrospect we would, with the very same evidence): This is the genuine linguistic phenomenon largely overlooked prior to the Pragmatic Encroachment literature, which covers indeed largely novel data; but this is a Pragmatic phenomenon: The charged unnecessary conflation/distortion is between the semantics of ‘know’ vs. the use of ‘know’, with the latter bringing out Pragmatic elements absent in the former. The very resort to the use of 'know' in such cases is in order to convey implicit Pragmatic messages given non-semantic/epistemic normative pressures.
Elsewhere I offered counter-examples to Epistemic Contextualism; see my (2012c).
A main thrust of (non-contextualist) Pragmatic-Encroachment in general upholds, in many cases, the talk that in high stakes we don’t know whereas in low stakes we do know, not only intuitively but also literally. On the Pragmatics-based treatment I offer here and elsewhere, this is to construe the intuitive usage mistakenly by mis-construing Sayability (use) intuitions as semantic—as transparently indicating semantic features, thereby erring about semantic transparency. The latter is of course not warranted when implicit contents are (Pragmatically) conveyed. Thus, the response is that what’s intuitive is the Sayability or Assertibility of such Knowledge-Ascriptions and especially their denials, whereas the Quick Jump Fallacy blocks the general inference from the proper premise ‘p is sayable/assertible’, construed as ‘intuitively sayable/assertible’, to the conclusion that p is true, where ‘sayable’ means ‘appropriate to say (in the situation)’. Indeed, such sayability indicates that what the speaker says is ‘right’ (in the situation of the saying), which is a characterization of the verbal-act rather than of the explicit content. But in Action-Directed contexts, to which bank-type cases belong, where the conveyed content/message specifies a steered-to action, non-sayability or sayability of the negation of a knowledge-ascription should be construed not in terms of the Truth-Value (of ‘p’) but in terms of whether the action in question (the ‘impending’ action) is the right action to take, i.e. (usually), whether it is right to act on p. Construing ‘what the speaker said is intuitively right’ (or even ‘intuitively true’) as ‘what she said is sayable’ (viz., appropriate to be said in the circumstance)—rather than as ‘intuitively, what she said is true’—is a main Pragmatics-based move: The move of reconstruing ‘right’ not in terms of truth but in terms of appropriate sayability, which in such contexts commits to its being right-to-act-on (to approving the steered-to action, where so steering is the implicit Pragmatic content).
But in many other Pragmatic contexts (which are not Action-Directed) ‘sayability’ is still called for when the speech-act carries a Pragmatic content even when it can’t be construed as ‘right to act-on’ (since Pragmatics also covers other domains—see, e.g., my (2018c), sec. 11). In all such contexts, the conveyed, implicit (Pragmatic) contents are not propositional and thus can’t be construed as having a truth-value. One needs to bear in mind that the basic ingredient of Pragmatics—what is implicitly conveyed—is Steering Thrust, which underpins Pragmatic contents (and is not propositional, on our Steering-Thrust Pragmatics). Stanley’s SSI move is wrong in construing ‘intuitively right’ as ‘intuitively true in the situation’ rather than as ‘intuitively sayable in the situation’ or just ‘appropriate to say in it’. The right construal (in Pragmatic contexts) of ‘sayable’ (or appropriately ‘sayable’—though 'appropriately' is redundant) is Pragmatic, and thus requires a suitable Pragmatic account, such as the one I have proposed. For a more detailed treatment of the Quick Jump Fallacy (i.e., the move from ‘intuitively right’ or ‘intuitively true’ to just ‘true’), see my (2017); for a more detailed treatment of the intuition ‘In high we don’t know’ see my (2020c). For the Pragmatics in question, which is Steering-Thrust Pragmatics, and in particular, in our type of cases, Action-Directed Pragmatics, see my (2018b, 2018c, 2018d).
That is, in contexts with only (or almost only) epistemic pressures, saying ‘It might be that the bank is closed on Saturday’ (in the bank case, with no action in mind), there need not be any Pragmatic content other than epistemic (which is, roughly, Steering Thrust towards an epistemic position). In low stakes, without the high risk, saying it might be appropriately dismissed, whereas in high stakes it may well be appropriately said and assented to. The Pragmatics-based treatment I offer here (and elsewhere) allows for such a locution (‘It might be that not-p’) to function as a Pragmatic operator without assuming that it has a truth-value (as a Pragmatic operator): What varies between Low and High is its Sayability. In High stakes, the speaker posts herself as taking seriously that contingency and not dismissing it as well as being against acting on p, while conveying con Steering Thrust (against it). Semantic treatments such as broad Pragmatic Encroachment (regarding ‘know’) are ad hoc in that they focus on ‘know’ and offer a treatment in terms of many knowledge-relations. Accordingly, such treatments are limited to ‘know’-idioms and would not be naturally and convincingly extendable in that way regarding various similar locutions (such as so-called alternative Skad-like locutions—see below). But locutions different from ‘I-don’t-know-that-p’ such as ‘It might be that not-p’ can generate the same type of puzzle under consideration since they have pretty much the same Pragmatic function without naturally inviting, it seems, Pragmatic-Encroachment treatment. Consequently, broad-Pragmatic-Encroachment accounts don’t offer a natural treatment for the puzzle in its more general form (i.e., covering also locutions like ‘It might be that not-p’ and not just ‘I now don’t know that p’, which function in the same way Pragmatically, including more complex cases such as embedding such operators in the scope of, e.g., 'I now don't think that ...', or 'now I can't say that ...' which, in our type of cases, are risk-dependent, steering towards not settling on the p-scenario). Accordingly, they come out as considerably ad hoc and not naturally generalizable, which is a significant strike against them. The significant cost in shifting from one knowledge-relation to many multiplies when we realize that there are various idioms that function (Pragmatically) in the same or similar ways in many cases but cannot naturally be treated similarly, and are thus left unaccounted for by Pragmatic Encroachment, which leaves the corresponding puzzle regarding them intact, although regarding formulations in terms of different locutions. By contrast, a Pragmatic treatment such as the one offered here yields that saying such locutions doesn’t convey a content with a truth-value but instead conveys Action-Directed-content/Steering-Thrust (in AD-cases) and with non-dismissive Posting (regarding the previously settled on contingency, and with a much weaker commitment to an Epistemic Position; see my (2021c)). For more on Posting, see my (2015e).
Of course, a non-Pragmatic-Encroachment approach such as my Pragmatics-based approach has to account for our intuition that what the husband says is right, as well as for related x-phi outcomes. As in general, the Pragmatics-based account makes the basic move that such intuitions are responsive to the sayability of what the speaker says, viz., to whether he is in a position to say it, and (in a Pragmatic context, which is usually the case) not only or only partially or not at all to their Truth-Value (thereby heeding the ‘Quick Jump Fallacy’ (see my (2017)). If we sympathize with his situation and welfare, such an intuition would be strengthened, and in particular if we also agree with his Steering Thrust (in this case, to his preferred action). Being Sayable or Sayability amount to being suitable/appropriate to say under the circumstances, which depends to a large extent on what the operative normative pressures are. (If only epistemic norms are operative, which is usually or at least very often not the case, the Steering Thrust is usually towards an Epistemic state.) Of course, the husband in the example does respond to the pressure of instrumental norms (manifested by the high risk), in contrast to the disinterested responder in the x-phi experiment who is practically unconstrained instrumentally and therefore responds in a way that reflects just her epistemic competence, which in turn renders her response as functioning indicatively (from her epistemic perspective) regarding her pertinent epistemic position, and accordingly vis-à-vis what she takes to be the truth-value of what is said: Since she assents, responds affirmatively, to questions posed to her about the vignette (e.g.: Does he know in the second stage that the bank is open?) she hardly conveys Steering Thrust: She pretty much only posts herself as being in a certain Epistemic state. This difference in operative norms between the two stages of a bank-type case accounts for the difference in the husband's responses, both of which we deem appropriate. (Accounting for our response when reading the example involves the phenomenon of ‘vicarious conversational-participants’, as I called it, who feel ‘engaged’ in the case and thus act ‘as if’ they are conversational-participants in the scene in real time and consequently under the pressure of the instrumental norms operative in the scene. I develop this account in detail (and in particular the important and interesting Audience Norm of Saying) in my (2012d) and (2019a)).
But even further (see previous footnote): The puzzle still persists regarding how we can view the husband’s response as intuitively true despite such findings (as in x-phi experiments). And some would say that we would continue to view it as intuitively true even if we were informed of such findings. But responses of this sort, if taken literally, are mistaken, and prone to fall prey to the Quick Jump Fallacy (see my (2017)).
I don’t have the space to consider embedding moves, e.g., whether the husband doesn’t know that he knows, or whether he knows that he thinks he doesn't know. In addition, elsewhere I argue that another important angle of how to view the puzzle is in terms of what I called ‘The Risk Bias’; for details, see my (2015b, c).
Or perhaps even lying.
And yet, we need to bear in mind the response available to Epistemic Contextualism (i.e., that the responder in the x-phi experiment was an ascriber in Low Stakes). But this doesn’t fully help SSI and Shifty Epistemology if what matters to them are the stakes of the subject (which are high) in her situation (which is for them the situation of evaluation, which is the one that counts) when she says ‘I don’t know’, which thus renders (on these approaches) that she doesn’t know (in her situation—then and there): We can consider an x-phi responder physically close to the subject in real time, who only cares about doing his job, and doesn’t care about the subject’s dilemma. She would then naturally respond as she would in a retrospective variation, viz., that the subject does know. So regarding the subject’s situation, which is for him the situation of evaluation, an indifferent x-phi responder can still sustain her response—that the subject does know. There doesn’t seem to be a semantic route to distinguish what the husband says in real-time about himself and what the x-phi responder would say about him in his very situation (given neither-contextualism nor Steady Pragmatics—see below). But of course, the subject is in high stakes in that very situation and the x-phi reporter knows it, although she is in low stakes. So Epistemic Contextualism and Steady Pragmatics can handle well the Ignorant High Stakes case, but it's a problem for SSI and Shifty Epistemology (that requires resorting to strong measures such as compromising the epistemic status of 'know').
If what matters to SSI regarding whether the husband knows is whether the subject, in our case the husband himself, who is said to know or not to know, is in high stakes (i.e., for him in that situation, then), then how are we to explain an indifferent x-phi responder’s response in that very same situation (i.e., that the subject does know)? An x-phi indifferent responder’s response, even if in the same situation, denying that the husband knows, would not be intuitively acceptable. So the indifferent x-phi response in the same situation that the husband is in (in real time) seems to be a counter-example to SSI and Shifty Epistemology if the relevant stakes are the stakes of the purported knower in his situation (in real time, which is the situation that counts for determining the truth value of the knowledge ascription). But neither proponents of Epistemic Contextualism nor me on my Pragmatic approach would have a difficulty explaining the retrospective knowledge-ascription (that the husband knew)—made after the dust has settled and the deliberation has faded and is no longer ‘alive’; yet not so for SSI or Shifty Epistemology, it seems (see my (2012a) and Blome-Tillmann (2020) as well as his earlier writings).
On my account, the explanation of the different responses is that in the cases when it would be ‘right’ to say that he knows, the dominant normative pressure operative on the speaker then is just epistemic and very low, and thus the literal content is conveyed (which is true), whereas if those normative operative pressures are primarily instrumental and high (due to the risk), the speaker, if susceptible to them, would convey an implicit AD-content (to the effect that they should deposit now). The onset of a considerable non-epistemic norm makes it possible to convey (in particular, a conflicting) Pragmatic implicit messages (to the previous one).
Note that Stanley also won’t be able (at least without adopting certain modifications) to distinguish x-phi responders who affirm knowledge when confronted with the vignette from what I called vicarious conversational participants, who would deny the knowledge-ascription, concurring with the subject (and primarily feel and act as if they are subject to the same normative pressures). For them, say, in a retrospective position, the stakes then aren’t high anymore since the outcomes are in the past. But their case would be different if they identify with the husband (in the past case) so that they thereby become susceptible to his earlier stakes even though they aren't, and can't be, their own direct stakes at their point in time. On my Pragmatics approach (as well as, I suppose, for Epistemic-Contextualism proponents), one could say that since they ‘identify’ with the subject, they adopt ‘conferred (‘as if’) stakes’, at least temporarily; or that they care about him so much that, at least for the moment, his stakes confer to them similar (‘as if’) stakes to his. The responder in the x-phi experiment isn’t a ‘vicarious conversational participant’, in particular since she doesn’t ‘adopt’ the subject’s stakes, and the instrumental normative pressures that permeate the husband’s case aren’t operative in particular for her in her retrospective situation.
Note too that Stakes' being high should be relativized to contingencies, e.g., as invoking risks associated with the contingency specified in the content-clause of the knowledge ascription in question: The husband can say (on Friday) that he doesn’t know that the bank is open on Saturday, and yet can say he does know that today (the day they speak) is Friday–since, given Steering Thrust Pragmatics, in the first he would thereby convey AD-Steering Thrust (where the operative normative pressure is AD) but not so in the second: This latter explicit message doesn't trigger any 'impending' action since it doesn't trigger an 'impending' Practical Inference -- unlike the former which does 'bring to mind' such an impending Practical Inference. But content-clauses specifying explicitly that there is such a risk can be employed, with an implicit AD content steering against acting on the complementary contingency (even if previously it has been epistemically 'settled on')—e.g., that he knows that there is a considerable risk if the bank is not open on Saturday? Intuitively, this is sayable, and this is what he (as well as we) may well say. Or else he can invoke the Skad and say (and we would concur) that he doesn’t know that the bank is open on Saturday. But the husband’s stakes are high in the situation in question, and there is no mechanism regarding stakes on the Pragmatic Encroachment approach for telling (after the news about the mortgage) that his second self-knowledge ascription that he knows that today is Friday) is sayable (by him, then) (or that this self-knowledge ascription is true, or is intuitive) and as is a putative self-knowledge ascription by him then that it’s risky if the bank isn’t open then and it’s not the case that he knows that the bank is open on Saturday! Given Pragmatic Encroachment, if the stakes are high, the subject doesn't know (other things being equal, at least things related to the pertinent stakes' being high such as the first two): So Pragmatic Encroachment should at least tell us when (at least, I would amicably suggest) a related self-knowledge ascription is true even when the stakes are high. Pragmatic Encroachment proponents have to deal with this issue—how to tell when a subject in high stakes does know that p? We just saw that his having high stakes about p lumps his (and our) knowledge ascription with the two content clauses together, whereas intuitively we’d say that he doesn’t know that it’s open but he does know that this is risky. At least, I would make a friendly suggestion: one should distinguish knowledge-ascriptions which are affected by the risk from those that aren't. On the Pragmatic approach I offer, the distinction would be between ones the Sayability of which is so affected vs. ones where it isn't. On this approach, he can say that he knows that there is a serious risk if the bank is closed on Saturday since in this AD context saying that triggers an 'impending' Practical Reasoning against delaying the deposit, and thus towards Depositing Now (rather than on Saturday).
So, on a Pragmatics approach such as mine, pulling these apart is easy: He does know both (i.e., that it’s open and that it’s risky), but he says he doesn’t know that it’s open (and is entitled to say that) since he thereby conveys an implicit Pragmatic message against taking the action in question (viz., to deposit now) since ‘I don’t know’ in a Pragmatic context—with high normative pressures—funnctions as a Pragmatic operator, which renders the literal content ‘idle’—i.e., not (at least not fully) conveyed. When he says as above that he knows it’s risky, his steering thrust is towards the action of not delaying the deposit, but the literal content is also conveyed since the two normative pressures align. He can say he knows today is Friday since, even in that context, there is no (obvious) action-directed implicit thrust, and thus (pretty much) only the literal content is conveyed. He is in a position to say he doesn’t know when in fact he does when ‘I don’t know’ functions as a Pragmatic operator in a Pragmatic context with high normative pressures that undermine the literal content from being (at least fully) conveyed, and when his saying it undermines the content clause's functioning as a premise in a Practical Inference leading to one of the actions ‘on the table’ then and there, which is not the action he prefers. This is how he would thereby convey (in an AD-Pragmatic context) AD-Steering Thrust against that action. Parenthetically note that his so saying that he doesn't know is undermines rather than acknowledges and assents to the alluded to Practical Inference: With the Skad in the first premise, its use is Pragmatic rather than epistemic whereas it's the latter the use of which is suitable for a premise in a viable Practical Inference, but with the former as a first premise the latter Practical Inference is derailed and rendered spurious and instead functions as a faulty justification for an already decided upon action—decided against (see also below).
Note that the core distinction here is between normative pressures (e.g., risks) in the context that bind the speaker/ascriber's pertinent verbal actions—that he needs to be responsive to—and those which don't, specifically that give rise to and motivate his corresponding conferred Steering-Thrust Pragmatic use and the implicit messages he conveys: The agent who counts primarily is the speaker since an account of 'what's intuitive to say' (specifically regarding knowledge-ascriptions and their denials) is vis-a-vis the linguistic, specifically Steering-Thrust Pragmatic, of the ascriber/speaker, not a speaker who doesn't engage in the pertinent verbal behavior that needs to be construed. The above normative pressures (that bind the speaker) I characterize for brevity as 'applicable normative pressures'. A main component of the above critique of Pragmatic Encroachment can be considered to be how to distinguish between risks that are relevant to the use of 'know' -- i.e., vis-a-vis the actions associated with saying the knowledge-ascription or its denial, and those that aren't. But proponents of Pragmatic Encroachment need to make such a distinction with an associated considered account for it. The obvious candidate is for them to say that a subject doesn't know that p if she shouldn't act on p, since then she doesn't 'know enough' in order to act on p. But this should be fleshed out in non-Pragmatic terms. On my Pragmatic approach, given risk is 'relevant' to a given verbal-act of the speaker/ascriber if it affects his pertinent Pragmatic use—e.g., if it binds his conveyed AD-implicit messages that are responsive to the risk under consideration.
Stanley raised a somewhat related objection regarding past-tense claims; but his response is inadequate (see Stanley (2005), ch. 11, (2), p. 106.): The x-phi responder would say the same even if presented with a vignette with the same story (even if just imagined) but as taking place ‘now’ for him (i.e., at the same time he considers it)—so long as he doesn’t become a ‘vicarious conversational participant’—doesn’t ‘adopt’ (at least temporarily) the subject’s stakes (some x-phi responders might, though I suspect most wouldn’t; see my (2020c)). The root of the problem is that Stanley's framework doesn't have the resources to distinguish between responders other than the subject who adopt the subject's stakes (to a sufficiently high degree, such as his family or friends) and those who don't (such as x-phi responders). The more precise differentiation is couched in my Steering-Thrust Pragmatics: Those who, by what they would say, steer towards the action under consideration (as the subject himself) and those who wouldn't—which is rooted in whether or not they convey the requisite Pragmatic implicit content in their use of 'know'. Trapped in his semantic approach, Stanley doesn't have these Steering-Thrust Pragmatics resources to make the right differentiation. DeRose does better, since he relativizes to the ascriber; but still, the tool available to him is ascriber's sympathy (within a semantic approach), which is a mere imprecise approximation: The core issue is what Steering-Thrust content (here, specifically, AD) ascribers in fact convey, not heuristic characterizations regarding their psychology (what would motivate them). The answer to this issue on my Pragmatic account is that, although the husband did know (in real time), he was in a position to say that he didn’t know— saying so was sayable for him (at the time, in view of the dominant instrumental normative pressures then) since his saying so conveyed an AD-steering message with the ‘I don't know’ operator functioning as a mere Pragmatic operator so that, under such high operative pressures, the explicit content of this utterance of his wasn't conveyed. But the x-phi responder, in her context with merely low and just epistemic normative pressures applicable to her (even if only from her perspective), was not in the business of conveying AD-messages to anybody, and thus didn’t convey an implicit content in what she said. (Note that different subjects might 'recognize' differentially certain operative normative pressures as applicable to them then -- might differentially 'accommodate' them (e.g., AD pressures). This need not usually call for a distinction of distinct idiolectal contexts.
This response is indeed related to DeRose’s ‘warranted assertibility’; but first, ‘warranted’ needs to be considered here Pragmatically, not epistemically; and second, what is considered (by him and others, in such cases and elsewhere) to be ‘asserted’ need not, I hold, be assertible—my intuition is that the husband is not, in the bank case (stage II), in a position to assert what he says (his denial of the knowledge-ascription) although he is in a position to say it (asserting need not be appropriate when there is an agreement on the steered-to action between the parties). But third, the weightier response is that in order to appeal to a satisfactory applicable Pragmatic warrant (rather than only to an epistemic one) one must have a satisfactory Pragmatic account to fall back on, pertinent to the type of cases under consideration (i.e., in my view, action-directed cases), which Grice, I hold, didn’t provide (Brown and Pritchard notwithstanding). Yet I have presented one—my Action-Directed Pragmatics [see my (2018a, b, c, d)]. DeRose rejected warranted assertibility construals of what the husband said, and thereby deprived himself of Pragmatic treatments. But again, the Pragmatics available to him then was indeed inadequate to handle such cases.
And in particular regarding his saying contradictory things while being in pertinently the same factual circumstances.
As is reflected in recent literature (see, e.g., Kim (2016)).
In some sense he is in a position to tell that it's not the case that he doesn't know. I will not expand on this here, but note that a competent speaker might well be in a position to tell what he would tell in a different context. And in a corresponding factual case but without the normative instrumental pressures (i.e., the risk), he would say that he knows that p (or at least that p).
Or, alternatively, to follow a different route regarding the conveyed message—not purely Pragmatic, but that of a shortcut: There is a standard construal that is not literal on which ‘I now don’t know that p’ functions as a common shortcut for something like the more near-sophisticated ‘I now am not in a position to tell/say that p’. So this is the option of that the speaker be so construed as conveying such a message. (Of course, it has been well recognized that the common locution ‘what he said’ can be construed in different ways, but I offer here a different construal specific to denials of knowledge-ascriptions and their kin.) I pursue this direction in my (2015b, c).
Which is how Stanley used it, as well as Shifty-Epistemology theorists. (Note that 'practical' appeals to instrumental concerns.)
I will not rely in this paper on the Action-Directed Pragmatics I developed and have as a backup for my Pragmatic claims; see my formal Pragmatics paper (2018b), where the conception is developed. It is presented informally from different perspectives in my (2018c) and (2020a). Again, I employ the adjective ‘Pragmatic’ with a capital ‘P’ as specific to the Pragmatics-based use (as distinct from, in particular, a ‘practical’-like use). I have so employed it in this paper (up to now and below).
In the sense of practical-like, related to stakes.
The resort to Gricean maxims, which are often not strong enough or targeted as needed or with a sufficient scope, won’t do to resolve the puzzle; see Pritchard (2005), Brown (2005). I am not comparing here the scope of Pragmatics and of 'being practical' (but bear in mind that ‘practical’ often is associated largely with ‘instrumentally-minded’, whereas a Pragmatics use can be induced by any significant normative pressure.). Shifting to increasingly higher-stakes case can trigger the viability of a more prominent Pragmatic use (conveyed implicit contents, and even only them—with the semantic content being suppressed).
Even after we take into account the outcomes of the relevant x-phi experiments.
Traditional, limited-scope, Pragmatic explanations by way of exploiting some Gricean maxim often miss the main, highly prevalent, Pragmatic phenomenon in question, which isn't limited to rising stakes being prone to alter our responses and to what we would say in seemingly denying or ascribing knowledge. This is since such puzzles trigger Pragmatic phenomena not focused on and attended to, or inadequately and insufficiently attended to, by the Gricean tradition.
The scope of this paper doesn’t allow to elaborate further, but hopefully what is presented is sufficient for an initial appreciation; but see my formal Action-Directed Pragmatics (2018b), and an informal presentation in my (2018c). In the last section of (2018c) I explain how this approach is to be extended to the traditional domain of Informational Pragmatics.
In this paper I can do only so much, and briefly. For greater detail, see my (2018c, 2020a) and my (2012a). For the bulk of the Pragmatic phenomena brought out by this sort of case, glimpsed at in footnote 8 in Sect. 1, but otherwise untouched in this paper, see my (2012a, d, 2015a). For the case that verbal-acts in Ethics are Pragmatic in character, similar in particular in view of their conveyed contents to those in Pragmatic branches such as Attitudinal or Valuational Pragmatics since the former are Steering Thrusts, see my (2020b).
Rather than a mere change in ‘standards’ (which is a contextualist move): A new normative pressure, such as the one induced by a new onerous risk, can have such a Pragmatic impact (even when via a prior change in Preferred Action) that doesn’t amount to a change in standards but which amounts to a new implicit content being conveyed by saying a given utterance (that is now sayable, though it need not have been so previously). Of course, if a different normative pressure comes into play, such as a new Instrumental pressure, or more specifically, in our case, in the form of, e.g., a new risk, conferring a new Action-Directed pressure, the actions under consideration project in turn new estimated prospects and risks in view of such a new normative pressure (be it instrumental or e.g., ethical, due to, say, an instrumental risk or an ethical risk). The resort to an invoked ‘new standard’ in such a case may be to a more demanding standard for performing the task at hand, e.g., epistemically (for decision-making), such as requiring more double-checking, better lab results, etc. But this would not be a suitable notion to employ in general in the light of the much broader perspective presented here: In our case a different normative pressure or one with a higher degree is now operative, with Pragmatic repercussions (on conveyed implicit contents and their dominance). The notion of standards here alludes to something like a purported higher requisite degree of epistemic position or epistemic confidence in view of a new high risk for performing an action that has been favored prior to the introduction of the new risk. But such ‘standards’ are a mere local way of considering different ways of bracketing degrees of a particular component of a relatively primitive decision heuristic (a correlate of subjective probabilities) in the service of primarily affecting the outcome of a decision (vis-à-vis actions under consideration, i.e., what to do now): The functioning bite of the new ‘standard’ is the standard of epistemic confidence requisite to perform a given action under the new circumstances although not a new standard for what now would count as knowledge (and not even a new standard for the Sayability of the pertinent knowledge-ascription or its denial). This standard-oriented approach thus merely extracts one way of ‘bracketing’ a particular feature in a specific appraisal-method in a decision-heuristic (in our case) vis-à-vis a specific action, and thus misses the new normative Pragmatic perspective with its by-far broader scope, and instead focuses on the subject's deliberation and certain cognitive features of it which are outside the realm of the requisite linguistic analysis here. Different degrees of epistemic positions (corresponding, at least partially, to different epistemic 'standards') would be called for on a fairly 'dense' scale' -- specific to the degree of the risk involved; but they wouldn't confer an almost 'dense' gradations of sub-senses of 'know': Such a profuse inflation of sub-senses is not called for in ordinary language, and there doesn't seem to be a natural carving that would reduce the huge number of different standards. Ultimately, the broader perspective involves changing normative pressures and how they interface in the Overall Sayability function (see below) after a new action-decision has been made, whereas standards-talk might presumably supply candidates for fitting a particular epistemic re-assessment purportedly needed in view of the new instrumental weight that should be reflected in the new ‘saturated’ Overall Sayability function. But my main point here is that such a purported epistemic re-assessment is not reflected by the use of ‘know’: The latter partakes in the resulting Steering Thrust away from the prior action-decision. A Pragmatic perspective of the case, such as the one offered here, which is called for anyhow, regardless of particular considerations regarding the use of just ‘know’ in high-stakes cases, renders otiose the narrow semantic straight-jacket that results in ‘new standards’ and a purported new inflated number of degrees of knowledge. Further: Such a standards-talk and different knowledge relations reflect a move of fitting purported degrees of knowledge as a component in a qualitative, Common-Sense (even if advanced) but merely purported decision heuristic, which is spurious since it doesn't reflect the speaker's actual decision making regarding which action to take, and invokes the correlate of something like a coarse-grained notion of well-founded ‘subjective probabilities’ in a decision-heuristic framework which has no use for a notion of ‘know’ whatsoever, whereas the Pragmatic perspective above presents such cases as involving merely signaling via (in our case, AD) implicit contents, without transparency regarding the underpinning decision to change the preferred action. The fundamental fact is that no appropriate change of set epistemic standards underpins the shifted use of know, nor even in the subject's own decision making. Rather, the picture I attempt to establish is that, in cases such as ours (as in an AD case confronted with a new risk), the subject, once provided with new information (as about a risk), re-assesses at a preliminary stage his action-decision well prior to his verbal behavior (that is our focus) which is an effect of it. The change in preferred action confers a new suitable verbal communicative steering of his interlocutor towards his new preferred action. So in the use of the denial of the knowledge ascription there is no change of standard—what is communicated is just (sometimes different) steering thrust towards an action (sometimes new). Even in the deliberation itself, outside our Pragmatic analysis, there is no change of standard. Rather, when performed as a 'qualitative primitive shadow' in advanced Common Sense of something like Expected Utility, no new 'standard' is invoked but rather a new rough qualitative assessment of the new risk vs. the expected gain (each in its corresponding course, in cases such as ours). The use of the Skad as a premise in a purported reconstruction of the subject's deliberation rather reflects a (prior) justification of a change in the decision—a rejection of the simple Practical Inference that has underpinned the prior decision (especially when 'now' is being used), which accordingly can be taken to motivate the/an alternative action -- although fundamentally its a Pragmatic use of an implicit AD content. Keep in mind that the resort to putative new standards has been a main lever of central types of Epistemic Contextualism. (Note that here I address the standard classical moves of the use of different standards for knowledge reflecting different senses of knowledge. I don't address her the approach of using 'know sufficiently' as involving different degrees of knowledge with the same sense of 'know', which is not a semantic approach to the puzzle.)
Of course, semantic (and linguistic, more generally) norms are pretty much always operative, so I can spare mentioning them.
That is, including already operative norms but with a new force, or intensity.
If possible. Actors with just simple Common-Sense would not be able to switch to a more demanding 2 × 2 decision-matrix. Rationally, we don’t have a good answer for such actors; but they would be prone to follow and benefit from an adaptive, entrenched bias (the Risk Bias), though crude, through a mechanism that provides an entrenched mode of response to such a change; see my (2015c) and (2015d).
For greater detail on the simple Practical-Inference heuristic and other Common-Sense heuristics and a general conception of when to apply which heuristic (covering also artificial heuristics, i.e., not in advanced Common Sense), see my (2015d).
For our present purposes (although not generally), we here need to exclude Knowledge Attributions (and other factive attributions) so as not to beg the question. By ‘empirical facts’ (here and in general) I exclude chances and ethical features; the latter are not our concern here (so long as we don’t deal with Ethical Norms). So perhaps a suitable term would be: empirical facts- (i.e., minus knowledge-ascriptions and the like). The main controversial issue is then whether the Truth-Value of a knowledge-ascription changes when the normative pressures change even if non-Knowledge empirical facts, i.e., empirical facts-, remain intact. The notion of an epistemic position is widely applicable here: I use ‘epistemic position’, following what I take to be a common use (even when not explicitly specified), for an internal epistemic position, covering, very roughly, what’s available to the subject from her own first-person perspective, in particular empirical facts and evidence (including assessments of objective chances, in one guise or another). I allow it to include referential terms. (I thereby shove aside epistemic issues regarding reference; see, e.g., my (1993).) An important but different usage is that of external epistemic positions (thereby true), covering, e.g., relevant chances. This distinction is part of the important general rift between internal and external features (and mixtures thereof), which I discuss extensively in my (2008).
I explore and develop it elsewhere but ignore it here; see, again, my (2015d).
I thus assume that hardly any norm which is not epistemic or instrumental is operative in our case, or else, more realistically, that if operative, it’s operative with low strength and its pertinent impact is very low.
I don’t have the space to expand in this paper on the phenomenon of leeways and extended (shifts) of epistemic thresholds that the subject is committed to (and posts); see the expanded version in my (2020b). I also don’t have the space here to expand on the other related important Pragmatic phenomenon—of how the speaker posts herself; for an elaboration of that, see my (2015f).
Normative pressure may vary in degree and in types. Compare with cases where the stakes haven’t risen but have changed—as in changes regarding which norms apply the pressures, or when the normative pressures apply equally (with the same strength) but in different ways (including different people). This also applies to specific finer sub-normative pressures (e.g., risks). Since we are concerned with the interface of normative pressures and actions, specifically linguistic actions (as reflected in the, e.g., Overall Sayability function—see below), it’s not enough just to say that a particular pressure is ‘operative’: The Overall Sayability function characterizes a particular speaker, and for it, the pertinent operative pressures must be as applicable to her: as she deals and copes with them—with or without awareness, thereby reflecting their effects on her, which is what matters. Thus, the pertinent risks and stakes must be vis-a-vis a particular speaker/actor under consideration, since different ones would reflect different operative normative pressures as they impact them differently. Even if the subject focuses on effects on another person she cares about (one way or another), the pertinent effects of the normative pressures on her would be (partly) vis-a-vis their effects on the other person (as is the case with vicarious conversational participants—see below).
Such new appropriate decision-heuristics may not be available to certain types of subjects—e.g., to simple-minded, uneducated subjects, who then won’t be able to perform an Inferential Ascent. I don’t have the space to present these phenomena here; for greater detail, see my (2015d).
Given operative epistemic norms and instrumental pressures, I focus here on Action-Directed norms (which are operative with varying intensity) in view of the puzzles under consideration here—bank-type cases. But the type of phenomena that could be instantiated here is much broader, since different types of norms and normative pressures (and different sub-types) may interface with epistemic norms in similar or different ways, with even greater complexity, with more than one non-epistemic norm-type exerting pressure (e.g., ethical as well as instrumental, or sub-types thereof). Note that conflicting normative pressures might be exerted in a given situation— hence the need for the Overall Sayability function, and different norm-types and sub-norms might be applicable.
Such gauging might be with some awareness, to some degree or other, or with none, with the meta-decision very often below the level of awareness.
An Inferential Ascent to even, say, an advanced Common-Sense heuristic, employing a suitable, more sophisticated, and thus subject to a different terminology, need not be available to agents with just simple Common-Sense. I also allude here to the Cognitive Fallacy in cases of new high risks, which I called the Risk Bias; for a detailed discussion, see my (2015b) and (2015c).
That is, the contrast is between the scope and power of an adequate Pragmatics of steering (involving, in our case, steering to action by Pragmatic means) and the tool of raising the threshold of ‘know’ variably in suitable contexts. The latter is very local (limited to the seeming semantics of a single term, and specifically to cases in which others are to be steered to action (or to inaction), and thus also limited to one branch of Pragmatics—Action-Directed Pragmatics) and as such quite ad hoc. But when we realize that this provides us with a mere glimpse into a large phenomenon that covers lots of Pragmatic domains -- not just AD, where various terms, idioms and constructions enhance the degree of Steering Thrust and where this is the basic mechanism of conveying implicit contents, the erroneously considered local Pragmatic Encroachment into the semantics of 'know' becomes evident.
I group these two norm-types together, covering also semantic norms when I explicitly invoke epistemic norms.
I consider here Knowledge-Ascriptions (among various other ascriptions) that are factual (i.e., their content-clause is): In cases of very Low Stakes, the Pragmatic Encroachment issue often doesn’t arise (since the operative norms are not strong enough to require and enable 'leeway', let alone overriding). I use here and elsewhere ‘factual’ in the sense of empirical—as purporting to specify empirical facts. But note that occasionally I consider the empirical facts without knowledge-ascriptions (and their kin) in order not to prejudice the Pragmatic Encroachment issue (recall the use above of Empirical Facts-).
But note that in usual practical situations we are not governed by epistemic norms alone (at least over extended temporal stretches): Such a case would leave out main constraints in our life such as the need to attend to our survival and manage our environment, constraints that impose limits such as how much time, energy, or other resources we can afford to, or should, devote to a particular epistemic concern (let alone Ethical pressures). In general, epistemic goods need to be weighed together with others, e.g., instrumental ones, even if the latter are at a minimal or low level, such as sheer survival (at least over periods of time), where of course commensurability concerns also arise. Pressing survival issues thus often compete with epistemic goals, which is unavoidable for living creatures. In general, practically, there isn’t a case of a competent epistemic agent operating (through sufficiently long intervals) only under epistemic pressures. (This issue has been of concern also to Michael Hannon).
See below, Sect. 5, for different sub-norm-types of Sayability (e.g., epistemic Sayability).
Saying is a linguistic act. Messages that a speaker can transmit can be conveyed in different ways—explicit or implicit, in different manners (e.g., with different degrees of Steering Thrust or different Postings, or involving non-linguistic communication), and in various sub-cases. For a more detailed view, see my (2018c), Section 5; for Postings, see my (2021a).
A competent speaker might fail (locally) to realize that a certain norm is operative in a given situation. Then the operative norm would help specify how he should act and what he is in a position to say or convey, but would not describe (or partake in describing) how she in fact acts and what she in fact says or convey.
Outside Pragmatics, of course.
Often without awareness, or requisite conceptualization (in some cases).
Notable exceptions are familiar Gricean implicatures. Semantic competence is governed by semantic norms. But when no familiar Gricean implicatures seem to be present, the customary stance was to consider just semantic/epistemic norms as operative (at least implicitly, by many). As this paper brings out, there is a lot more to Pragmatics than Gricean Pragmatics or Gricean implicatures, and the Steering Thrust Pragmatics I have offered (and have in mind implicitly in discussing its repercussions, e.g., here, to bank-type puzzles) is very different from Gricean Pragmatics even when they are both applicable to a given verbal act (including cases where, in my view, Steering Thrust Pragmatics overrides Gricean or neo-Gricean Pragmatics); see my (2018b), (2018c), and (2020a). (So-called WAMs seem to invoke just Gricean Pragmatics, at least when they have been considered initially, which has been insufficient for backing them up. But they are also criticizable on various other scores, as has been pointed out in the literature, first by DeRose, whose critique I and others found unconvincing; for my response, see (2012e). But of course, at that time Gricean and neo-Gricean Pragmatics were what he and others could resort to.)
See notes 2, 3, and 4 above. The objection regarding warranted assertibility was raised early on, but was deflected (e.g., by DeRose—see his ‘WAM’.) The response that this was an unwarranted deflection of the presence of pertinent implicit messages was made in Brown 2006 and Pritchard 2005 (though not in these terms), but inadequately, in my opinion, due to insufficient back-up by Gricean Pragmatics on which they relied. The prevalent discussion of a Norm-of-Assertion focuses often, after Williamson, on an epistemic Norm of Assertion, which helps disguise implicit contents conveyed in the assertion itself: An epistemic Norm of Assertion would invoke only epistemic norms, and thereby epistemic commitments, such as to Truth and allegedly to Knowledge. But asserting conveys Steering Thrusts more strongly than mere sayings (i.e., weightier degree of Steering Thrusts), and then isn’t confined to merely a ‘passive epistemic commitment’ (which I would construe as a derivative of Posting): Asserting invokes steering (to an action, to an epistemic state, to an attitude or emotion, to an Ethical position, etc.) and is thus a main vehicle in Pragmatics. Such remarks seem to me to be consistent with Williamson’s position regarding the Norm-of-Assertion (although the position that an assertion commits only to its explicit content has been advocated by Goldberg, very much contrary to my position). Cast in my terminology, the epistemic dimension of assertion invoked a normative requirement of posting yourself as knowing that p (given Williamson's thesis) in addition to, often, steering towards the Pragmatic Stance of knowing that p—and/or, often, steering also towards another Pragmatic stance(s). For more on this thesis, merely touched upon here, see my (2021b); for more on Posting, see my 2021a). For More on Williamson’s Knowledge Norm of Assertion, see his (1996), (2000), Goldberg (2015) and the extensive literature on the subject.
I.e., for ‘p’ to be sayable is for it to be overall appropriate for the speaker to say it under the circumstances. ‘Overall’ covers all significant applicable norms, whereas ‘epistemic sayability’ covers sayability just under epistemic norms, and so on. Sayability comes in degrees: the Sayability of p comes in different strengths under different operative norms (as a first approximation, since they might not be retainable and thus not stable under the omission and addition of other norms, which undermines the strict viability of this phrase; for details, see my (2021b)).
Usually without awareness. Of course, we primarily track pertinent relatively likely candidates, particularly ones we attend to. (By ‘likely’ I mean: likely candidates for the speaker to use then).
This seems to involve a cognitive mechanism of the sort that Kahneman called System 1; see his (2011).
Again, of course, primarily likely ones that we take into consideration, to one degree or another. But ‘p’ may or may not be sayable by the speaker in the circumstances even if she doesn’t gauge it.
The subject of how the Overall Sayability function is determined, in terms of operative sub-norms (for a speaker, in a particular situation), is in my view empirical, and I can’t delve into it further here. Same vis-à-vis the issue of when and how operative sub-norms have a strength and polarity as a function of operative sub-sub-norms.
The requirement of being candid very often figures in operative norms.
By such a profile I have in mind the operative normative pressures and their strengths under the circumstances.
By the facts involved I mean (empirical) facts that are epistemically relevant to the Truth-Values of what’s being said as well as Pragmatically relevant—to the topic under consideration (such as impending actions). Yet note that invoking the ‘sayability in (such) a case under different norms’ may be empty in various cases since adding or removing norms as operative might be incompatible with keeping the empirical facts of the case intact (e.g., if a certain norm is invoked by certain facts, e.g., a certain risk: removing the instrumental pressure due to it may be incompatible with leaving the case (factually) as is; and same with respect to adding norms: Adding or removing normative pressures may well require modifying the facts of the case; see above, note 88.
Again, I take compliance with Semantic norms for granted here.
The conveyed/expressed content(s) of what he said, which are (as I take them to be) public features of his utterance, usually, but not always, coincide with what he aimed at conveying/expressing. I don’t dwell on cases when the speaker mis-spoke. Linguistic Pragmatics (which I take myself as engaged in), as I see it, deals only with the public features of verbal acts or utterances—not with what the speaker ‘intended to say’, if different. Gestures, intonation or facial expressions don’t figure in Linguistic Pragmatics (unless described, as in a written text), strictly speaking, as I take it to be and invoke here, although they are very prevalent in speech and oral communication but much less so in written texts, which is the purer form phenomena of Linguistic Pragmatics or their description. By ‘verbal’ I emphasize the linguistic character (whereas ‘speech’, literally, doesn’t cover written communication).
For further details, see my (2020a), Section 2.
I.e., a saying verbal-act, e.g., an assertion; and also, they may align or conflict vis-a-vis another action—not just a Saying; but in this paper, and specifically here, we focus only on Sayings (indicative verbal acts). So more generally, norms can align and conflict also vis-à-vis other speech-acts and communicative actions in general—not just linguistic-acts but also, say, gesturing or having a certain facial expression, or using a particular intonation, as when each of the various operative norms (say, instrumental and ethical) yields the same polarity for the Steering Thrust in question (a case of alignment), or when two conflicting operative norms (applicable to that saying) yield Steering Thrusts in conflicting directions (e.g., polarity '+' and polarity '−' vis-à-vis the action under consideration then). The outcome may be pro Steering Thrust (i.e., for performing the action, with a certain degree), or no Steering Thrust (if the two conflicting norms neutralize each other), or negative (con) Steering Thrust—Steering Thrust against performing the verbal act (or action), to a certain degree (of strength). Aligning and conflicting apply more generally to other Pragmatic Categories as well (i.e., to communicative acts falling under Pragmatic categories other than Informational Pragmatics or Action-Directed Pragmatics, e.g., Emotional Pragmatics or Attitudinal Pragmatics, expressed by other verbal acts or only by non-verbal reactions, such as a facial expression; but the latter wouldn’t fall under Linguistic Pragmatics).
I don't dwell here on the phenomenon of leeways (see below).
Including, which goes without saying, the context. I’ll often, for brevity, omit contextual relativity when it’s obvious. When I use ‘case’ or ‘situation’, I extend its use to include the (pertinent) conversational context.
If one normative pressure is weak (say, instrumental), and there is only one other significant normative pressure (say, epistemic) and it is strong, then the latter is dominant regarding sayability. For the sake of brevity, I will cover sub-norms and normative pressures when taking just about norms.
In a case where there are only two (significant) operative norms. If the two norms more or less balance each other (vis-a-vis a certain saying), there would be mutual canceling: there would be no normative constraints of these two types vis-a-vis the speaker's saying it or not saying it.
I don’t discuss here cases when almost only epistemic norms are operative. In general, over time, a scenario in which only such norms are operative is not realistic—see footnote 79. Usually epistemic norms require other operative norms for steering for or against non-linguistic actions.
Or more generally, via Action-Rationality’s norms, which cover epistemic norms. I will allow myself sometimes to blur the difference for simplicity sake.
I have in mind here a simple Practical Inference, which would have to be ‘viable’ for the sanctioning not to be prima facie.
In such a case, the actor should be in a position to ‘run through’ the Practical Inference, assertorically or sayingly. Such Practical Inferences are ‘dynamic’, and are subject to non-deductive viability (corresponding to validity in deductive reasoning; see my (2015d)). Epistemic norms sanction an action for a collective if this condition holds for enough ‘weighty speakers’ in the collective—such as (frequently) a large enough majority. For obvious lack of space, I don’t touch here the vast literature of the past two decades or so on Assertion and Action (and collective epistemology).
Or both sanction not performing it; or are both just conducive to performing it or to not performing it. To avoid cumbersome formulations I occasionally ignore such 'under a threshold' formulations.
Yet the notion of 'sanctioning', in particular about a verbal act, needs to be taken here with a certain grain of salt, and I allow myself a certain imprecision regarding it. In this paper I don't have the space to dwell on characteristics of the Overall Sayability function (see below), nor on thresholds for particular normative pressures, nor on cases where a particular normative pressure is taken into account even though it's below its threshold but allows a' leeway'. For a more extended treatment, see my (2021b).
I.e., saying that premise is prone to invoke a certain normative/instrumental/wish constraint that is ‘impending’ (in the sense it's that the speaker in question at that situation ‘has that wish/concern in mind' or that it's prone (for him) 'to come to mind’). Once the speaker has said the first premise, that ‘wish-like’ constraint is invoked or ‘selected’, given the hearer's perspective. This is typical when a subject reasons his way towards an action. Of course, from the speaker's perspective, her wish is operative first, since on the basis of it she selects what to say. This way of looking at things reflects my outlook on how Pragmatics should be formally modeled; see my (2018b), (2018c). On the other hand, as in the case of a Skad, the speaker might have that wish/constraint in mind first, and on the basis of it selects what to say, in which case the saying steers to action without suggesting a motivating reasoning -- it only seems to do so, and it would be a mistake to consider it as providing a reason or reasoning (other than the speaker's say-so).
That is, an instrumental pressure in the first situation (say, the first stage of the bank-case) can be overridden by a new instrumental pressure introduced in the second stage (by the new risk), yielding also that epistemic normative pressures are shoved aside (vis-à-vis overall sayability), so that what was not sayable before is sayable now (e.g., ‘I now don’t know that the bank is open on Saturday’). Such a reversal of sayability of a major premise (e.g., ‘I know the bank is open’) undermines the viability of the Practical Inference employed in the first stage. Such a retraction of its overall sayability seems to sanction a new seeming Practical Inference (leading to the new Preferred Action, but which is no longer a simple Practical Inference, yet is still in Common Sense). But since the original premise has been retracted (in a Skad form—see below), i.e., it's not longer overall sayable (though still is epistemically sayable), the overall sayability of the new premise now need not reflect epistemic sayability: In the bank case, the sayability by the husband in stage II of the Skad (i.e.: Now I don’t know that p) doesn’t convey its explicit content but just Action-Directed Steering Thrust.
One can now consider a candidate for a Practical Inference employed now with the sayable Skad as its first premise, with the high risk as a second premise, a ‘wish’ premise of being unwilling to face that risk, and with a comparative premise that assesses the avoidance of the risk as far more weighty than standing in a long line. But this is not a simple Practical Inference—it is complex, invoking risks and the comparison of at least qualitative, very rough, utilities. Furthermore, it’s not a Practical Inference at all, since the sayability of (endorsing, ‘running through’) the first premise doesn’t convey its explicit content and as such doesn't convey empirical information: it conveys AD-Steering Thrust that is the presumed conclusion of the seeming Practical Inference. Therefore, saying the Skad need not, and in our case, does not, represent the reasoning of the speaker: It only conveys an Action-Directed Steering Thrust (and a certain Posting of a compromised epistemic commitment (by ‘extended leeway) as well as posting of a retraction of a preferred action (the 'now' in the Skad plays a role here)). Yet whether or not the presented overall-sayability of the premise reflects epistemic sayability, and to what strength, is not communicated by the speaker, explicitly or implicitly, and so the hearer is unable to extract epistemic (with a ‘factual’ content) sayability via the Skad with a definite strength even if she has full epistemic confidence in the speaker: The speaker doesn’t post himself (in saying the Skad) as having a certain epistemic position vis-a-vis p (beyond the bounds of the ‘extended leeway’). His saying the Skad conveys Steering Thrust away from the alluded-to action (of delaying the deposit) and undermines the prior alluded-to seeming Practical Inference (since now there is informational main premise). For the notions of leeway and extended leeway, see my (2015f); for the important Pragmatic component of Posting see my (2015e).
The saying is actionable (in the normative sense, not just the in the 'able to' sense) presumably if the Practical Inference in question is 'viable'.
Note that one can act on p even if the 'impending' Practical Inference is not viable: acting on p need not be appropriate. And acting on p would be ambiguous if there are two 'impending' 'wish'-premises leading to very different, let alone incompatible, actions, even if one underpinning Practical Inference is 'viable' but not the other. Furthermore, 'actionable' seems also to be ambiguous between 'can act on p' and 'can appropriately act on p'. So ‘acting on’ the premise (the saying) ‘I know that p’ in stage I would be to follow up on the AD-Steering Thrust conveyed by that saying. The action in question is, or becomes, ‘impending’ (already is 'on the table', or likely to come to mind on the basis of p). There is a Practical Inference with that first premise (of stage I), leading to a steered-to action via another ‘impending’ wish (the wish not to stand in a long line now). But ‘acting on’ the Skad is just to follow up on the Steering Thrust it conveys—this is all that it conveys, despite its expressing an epistemic state. But there is no Practical Inference that can represent a reasoning with the Skad as its main premise and the action in question as its conclusion, since the first premise doesn’t convey any descriptive content. How does a hearer figure out what action she is steered towards? A full-fledged unorthodox Pragmatics is needed for that; I have proposed my Steering Thrust Pragmatics; see my (2018b).
In addition, there need not be a single action specified by ‘acting on p’ also if there are different applicable normative pressures which allow for different candidates for supplementation for being different extra premises that yield different ‘impending’ Practical Inferences that lead to different actions (at least so long as no one normative pressure obviously trumps the other). Thus, two 'impending' extra premises might be an instrumental wish premise and an ethical premise that may well lead to different actions (via different Practical Reasonings).
Not only when two very different actions are candidates (with one ‘impending’ goal/wish), but also when two different goals/wishes are ‘impending’ or with two conflicting or 'diverging' different 'impending' premises—say, one being instrumental and the other specifying a relevant code (e.g., an ethical code). The locution ‘acting on p’ is employed in particular by Fantl & McGrath. But the locution accordingly need not specify a unique action if there is more than one impending Practical Inference (unless the use of that locution is relative to a particular Practical Inference). Accounts of different types of Practical Inferences and of what a viable Practical Inference amounts to are called for; for my account, see my (2015d).
Again, or sufficiently so. Each norm-type, such as the epistemic norm-type or the ethical norm-type, etc., may cover a variety of particular norms that fall under it (or even other sub-norm types).
Assume for simplicity just two operative normative pressures.
I.e., when the operative normative pressures are aligned vis-à-vis the action in question—as when both ‘support’ it. An epistemic normative pressure of p supports an action if there is an 'impending' viable Practical Inference that leads from the Overall Sayability of ‘p’ to the action in question—in which case acting on p is to perform that action. (But this need not be the only route—'acting on p' can vary with Pragmatic branches.) An important exception, very pertinent to our concerns here, are cases where the explicit content is not conveyed by saying that p: The viability of a 'dynamic' Practical Inference (in its standard (or simple) form) relies on the explicit contents of the premises (as well as on the degree of their Steering Thrusts). I elaborate on some more intricate ways of alignment and non-alignment below. For more on Pragmatic branches, see my (2021b).
And/or also a degree of Sayability. (I here simplify a more fine-grained treatment in my Formal Pragmatics for AD contexts, where I consider separately a binary function that yields polarity (as with or without Overall Sayability) and a function that yields polarized degrees; see my (2018b) and note 119 below.
One might make observations regarding what such a function looks like, but ultimately it’s empirical experiments that are needed to identify it; yet outcomes in various considered cases can confirm a qualitative picture.
More or less. The space of Operative Norm-type pressures would in general be an n-dimensional space, for n operative norm-type pressures. In a simple case, it’s a 2-dimensional space; and a case when the only dominant normative pressure is epistemic and the others are very minor is what we call a ‘pristine’ case.
Collective non-alignment requires some pair-wise non-alignment. More precisely, one may distinguish between a binary Overall Sayability function, yielding just Sayability/non-Sayability, and a degree-of-Overall-Sayability function, which yields the degree of Overall Sayability (which may be positive, negative or 0) (If one wants to dig deeper, one may consider it too as a function not just of a given saying but also of the degrees of sub-norm Sayability for each operative norm-type pressure. Such a degree of sub-norm Sayability may result from a single sub-norm pressure; but if there are a few, this function would again depend on the polarity and degrees of each component operative sub-normative pressure (and need not be 'accumulative').
I.e., below a pertinent threshold (if there is one).
We can thus say that alignment for (as opposed to against) Overall-Sayability amounts to positive-polarity-alignment for all operative norms (or netural, for some but not all). The alignment in question of the epistemic sayability is on the one hand of the epistemic Steering Thrust and the polarity of the Steering Thrust of the saying (i.e., of its overall sayability), which can be associated with the action in question (in our type of cases) or with other Pragmatic Stances., on the other.
For different operative normative or sub-normative pressures.
‘Do your job’ is a general norm, but what counts as doing your job varies with the specific job or position in question in certain circumstances in different communities.
I will not expand on perceived versus actual relevant operative normative pressures. To avoid further discussion, for lack of space, suppose we consider cases where they coincide; but see the next paragraph (specifically, the discussion about Ignorant High-Stakes).
Or when its intensity changes.
If prompted appropriately.
Or a significant change in the intensity of an already operative normative pressure.
In the sense of a subject’s being aware of high stakes or a high risk.
The insistence of Epistemic Contextualism on the ascriber’s perspective is analogous.
Of course, Pragmatic-Encroachment proponents would construe it differently—as the content conveyed being the literal (expressed) content, giving rise to, e.g., ‘know’ being used artificially (when construed from the opposite Pragmatic alternative presented here).
Even though it’s manifested in the shifted epistemic commitments of the speaker (the husband) — the epistemic position he posts himself as having, which I discuss elsewhere (see my (2015e, 2015f)). Note that Fantl and McGrath, like me, regardless of how diverse our accounts of the case are, shouldn’t have a problem with construing such exclaims as assertoric (unlike Goldberg). But unlike me, they construe the conveyed content as just the explicit content.
Especially when the dust has settled.
Even if gradually. I analyzed such a phenomenon in a similar way in my (2012). Note an analogous move by DeRose in his discussion of Skepticism—that the hard-core skeptical context ‘fades away’, i.e., the threshold drops gradually; compare also Blome-Tillmann (2020).
Accordingly, there is no longer (after the dust has settled) AD Steering-Thrust invoked (see below), particularly by pertinent knowledge-ascriptions or their denials. The ‘I know that’ and especially its denial usually perform the double duty of epistemic operators as well as Pragmatic operators—but mostly the first if the situation is ‘pristine’ (i.e., with non-epistemic normative pressures not being significantly operative).
Or at least predominantly.
No other significant norms need to be brought to bear in order to just recount a long-past/historical event, and often none is (with obvious exceptions, e.g., ethically ‘pregnant’ cases).
I take Action-Directed norms to provide guidance about what to do in given circumstances—including what to say or write (verbal acts). They can underpin utterances in present tense (more or less—as in a suggestive mood, but also in an imperative form) given a live dilemma or issue. I am not considering here an assessment of past actions or inactions in the light of ex-post-facto hindsight. (Of course, when such hindsight (ex-post-facto) information can be brought to bear, Action-Directed norms can be invoked in a derivative way—e.g., by way of learning from experience, as about what could or should have been done, and accordingly—as when ethical norms are operative—by determining responsibility, as in holding agents accountable, etc. (which are outside our interests here). But then no AD Steering Thrust is conveyed. Another type of case is one where the speaker operates verbally as if the normative operative pressures of the long-past dilemma are still operative—and, as such, functions as a vicarious conversational-participant (as if she is present in the actual past case), which is not the usual case; see my (2012d)).
Specifically, in the ‘traditional’ sense of epistemic, which is devoid of a pragmatic element (such as stakes).
Compare Blome-Tillmann (e.g., (2020) and (forthcoming)).
And this holds accordingly in a version in which the analogue of the second stage is an entirely new incident—when again considered in retrospect.
I ignore here issues such as knowing ‘implicitly’ or ‘latently’ in real time, or when confused and ‘taken out of balance’ temporarily as indeed can happen when ‘unprepared’ subjects are confronted with a very potent new risk, especially of a type that is new to them (e.g., a temporary confidence-loss). I discuss such issues in some detail in my (2015c). Very considerable genuine confidence-loss can generate an epistemic phenomenon that is present together with the Pragmatic phenomenon under consideration here, which I don’t address here. But in such a case some (factual) confidence decline is normal (even if not epistemically rational), and is present together with the Pragmatic phenomenon under consideration. Recall that, given a Pragmatic perspective, intuitions about truth-values (e.g., in 'That p is intuitively true’, or: It’s intuitively true that p) reflect not just semantic/epistemic competence but also a Pragmatic evaluation of sayability, which, as such, can bring to bear other pertinent normative pressures (if there are significant ones), as in the sense that p' is Sayable then and there by the subject (or the speaker).
Assuming that no such non-epistemic pressures are significantly operative at that time of the retrospective recounting. This is indeed the expected result in well-done x-phi vignettes when the incident on which the responder is asked to comment is told in past tense in case such a responder is not subject to the non-epistemic normative pressures that the conversational participants in the incident are subject to (nor, of course, to any other, which is the usual case). Even then the experimenter needs to control the respondents so as to secure that they don’t form an attachment (to the subject or some conversational participant, or a preference regarding the outcomes) due to a favoring stance that can affect their solicited intuitions and thus contaminate the latter as appropriate evidence for truth-values. Such controls need to be brought to bear especially when the story is presented in present-tense since then (unlike past-tense) it’s easier to imagine the outcomes as not yet fixed and the responder is subject to such potential biases even more. We can briefly explain why the former is a good test for checking semantic intuitions (specifically, about truth-values): When the case is cast in past tense so that the dust has ‘settled’ (on the incident, which is sufficiently earlier) since the pertinent decisions and their repercussions have already taken place, a speaker who is sufficiently familiar with the incident is not prone to assume the stance of a ‘vicarious conversational participant’: The action-decisions in the long-gone past (specifically, in our example, as to when to deposit the check) are no longer alive, and consequently, the AD-norms are no longer operative, and thus don’t interfere with epistemic (and semantic) operative pressures. When only the latter are operative, they reflect semantic features in relative isolation. For the broader and more complex phenomena that arise in such cases, see my “Audience Norm of Saying” (2012d).
In that form, ST-Pragmatics is no longer operative in the Action-Directed domain, though it is operative in the informational/epistemic domain—with Steering Thrust towards Epistemic Positions; see my (2018c), last section.
One might think that what the speaker (the husband, in such a case) says represents the conclusion of practical reasoning that underpins the speaker’s decision/action-reversal, which can be summarized as: He now thinks: I now don’t know that the bank is Open on Saturday. Hence, I can’t act on knowing it (as I concluded before); hence I am no longer in a position to decide to Delay the Deposit. But the point of much of our discussion is that the premise here (I now don't know ...) is not adequate for Practical Reasoning since its explicit content is not conveyed (and what's conveyed doesn't have a truth-value or probability). So such Practical Reasoning is flawed. The role of the premise here is not to serve as a premise for such reasoning of his or to represent a fact or a knowledge-ascription fact, but to steer his interlocutor—his wife—to his new preferred action. Epistemically, he is not in a position to convey the explicit content of the premise (viz., that he doesn’t know), and in saying it he doesn’t represent himself as being in that particular epistemic position, and so such an inference would be epistemically and thus rationally defective. (Recall that even in a case of Confidence Loss that is not a devastating one he would be in a position to retain his belief to that effect, even if he might not be in a position to tell whether or not he knows it, and thus wouldn’t be in an epistemic position to outright deny that he knows it, a-la the literal construal. (Of course, here I ignore the option of considering the Skad as a shortcut—see my (2015b) and (2015c).) True, he is now bent on acting as if he can’t tell whether he knows that the bank is open on Saturday—this is how he correctly presents himself (during that time interval). But this is not the bank case, since in the latter he doesn’t convey the literal content of what he says: Thus, in a type of case like the bank case, with a huge risk but not a serious confidence loss, to represent the first premise as an item in rational reasoning is a mistake: It’s unwarranted to so consider the standard bank case as a case of a massive confidence loss, which would then miss the dominant Pragmatic aspect of the standard version. The role of this seeming premise in the bank case is not to serve as a premise in rational Practical-Reasoning, but rather to steer others to the husband’s new preferred-action. (For a related point, see Baron Reed, (2013)).
For an account of reasoning in Common Sense, see my (2015d).)
Some of them or all of them. As noted, we have focused on the original framing of the case in terms of two stages and a retraction by the husband. But the same treatment, mutatis mutandis, holds also for the other variant of the case—where we have two distinct and unrelated cases corresponding to stage 1 and stage 2 respectively and therefore with no retraction, since the husband in stage 1 and the husband-correlate in stage 2 are distinct and unrelated individuals.
More precisely: Conversational Contexts in certain physical Setups. I called such an ordered pair (of a Conversational-Context and a Setup) a Setext; see my (2018c). For simplicity I don’t employ this terminology in this paper, and thus use the more lax and imprecise ‘context’ or ‘situation’ or ‘circumstances’ or ‘case’.
At least much of the time. Often the conveyed content would consist of both the explicit content as well as the Pragmatic content—Steering Thrust; but in some cases, as we have seen, it would consist predominantly of Steering Thrust.
And, in addition, rational-action norms, which I don’t focus on in this paper. In this paper I focus on the Pragmatic phenomenon, and take as ‘within a black box’ the subject's decision-making. When the husband hears about the risk, he reverses his action-decision. Only then, accordingly, he says what he does and steers accordingly. I thus don’t at all address how he reverses his action-decision. A predominant theme that I advocate here is that in what he says he conveys Pragmatically Steering Thrust to others how to act. But I also hold that he is opaque (to his conversational participants) about how he has made his decision. He steers the hearers to retract the prior simple Practical-Inference (if the hearers have endorsed it) in view of his retraction. But his retraction is merely seemingly of the explicit content of what he says (that he doesn't know): What he retracts is the sayability of the major premise, and he thereby steers his hearer towards retracting their prior action-disposition. But he is vague about what pertinent epistemic position he now occupies as well as about which content of the first premise he withdraws—the literal or the implicit or both: In fact he retracts just the Pragmatic implicit content— the Steering Thrust. But he can be seen as giving them a reason not about which epistemic position to hold about the explicit content of the first premise but only about how to act—to retract the prior action-decision (if they shared it) in view of his authority and this new stance of his; but he doesn't give them a reason to think that he now doesn't know: Rather, he now presents himself as a source with a certain AD-position, different than his previous position.
Given sufficient candor and expected transparency and without complex manipulation. Elsewhere I developed the twin Pragmatic component of Posting—in addition to conveyed Steering Thrust: A speaker Posts herself in steering via a verbal act as holding a certain Pragmatic stance -- of some sort) as being such-and-so (e.g., holding an epistemic position, but this is merely a very special case—a specific Pragmatic stance in the Informational domain of Steering-Thrust Pragmatics). So I find ‘signaling’ and ‘representing’ as too loaded with ‘aimed transmission’ for the Pragmatic stance of Posting, which is entirely passive: Like posting a ‘Dead-end’ sign in the beginning of a short alley, independently of whether this or that passer-by would heed the sign. I take signaling to usually involve conveying, although it need not. For details, see my (2015e). Posting is ‘passive’, with a variable degree of awareness (if any), in view of a commitment due to an operative norm, unlike conveying, which is ‘active’. There is Posting of a Pragmatic stance (such as an epistemic position), but also, in my opinion, Posting of Approval—of a certain action, or of having a certain epistemic position, or of an attitude, an emotion, an ethical stance, etc. Such a pertinent pinpointed Posted Approval (plus pertinent requisite Candor) reflects that the speaker holds that Pragmatic stance, such as willing to perform this action or having this attitude, etc. (if she is in a position to: she might be incapable of, e.g., performing the act, or of having that attitude, etc.)
Here I focus on simple Practical-Inference; see my (2015d).
I consider a mere verbal-act of saying—of factually spelling out a contingency—as indicating that the speaker has settled on it, and of asserting it as endorsing it.
Which would align with steering the listener to act on p under the circumstances if acting on p amounts to, and is understood as, performing the action in the conclusion. In many cases, the context (in particular, 'impending' wishes and the like) settles which action is selected by ‘acting on p’, especially when this is the action under consideration. But in various contexts this need not the case: There may be more than one action that so would fit ‘p’ as a premise (in an 'impending' Practical Inference, especially if the ‘want’ or ‘welfare’ of the silent extra-premise allows for more than one suitable candidate). So the notion of acting on p, when successfully picking up a certain action (in a context), hides unspecified processing which ‘otherwise’ need not be straightforward or specific of selecting such an action. In addition, one must be aware that sometimes ‘acting on p’ determines only a particular action (type) when a particular, more specific, action needs to be decided on (different more specific actions could fit). Further: acting on p need not signal a literal action -- it might reflect some Pragmatic stance (an epistemic position, an attitude, an ethical stance, ...) or several.
Accordingly, bear in mind the inadequacy (in general) of resorting to the straight-jacket of explicitly specified ‘steering-filters’ terms for classifying steering messages (e.g., ‘advising’, ‘instructing’, ‘directing’), employed (to a certain extent) by Gricean-style Pragmatics, and their poverty in natural language. The variety and subtleness of the spectrum of Steering Thrusts, in many cases in effable (not explicitly specifiable) and subject to variations of modes and degrees, is ignored and trampled over by such a conception of being artificially confined to the minimal menu offered by terms in natural language: A good deal of Steering Thrusts are conveyed partly or wholly non-explicitly—by intonation, facial expressions and gestures, which are often ineffable, as well as by terms which modulate the degree of steering (e.g., ‘know’—via its Pragmatic profile; and similarly for ‘suspect’ and ‘conjecture’). This is one place where Linguistic Pragmatics (my main concern) is seen as an integral part of a more comprehensive domain of human Communication.
I use here ‘signal’ whereas the more precise term is ‘post’. Posting, and in particular posting oneself as retracting, is part and parcel of Pragmatics I don’t have the space to expand on it here, as well as posting a preferred action and one's approval of it; see my (2021a).
The husband’s preferred action is clear and his wife too seems to prefer to Deposit Now.
This is clearly the case vis-a-vis the husband. The wife asks a question, and we don't deal here with how questions steer interlocutors. But when the wife asks ‘Do you know that the bank is open on Saturday?’, she positions herself as probably (or possibly) not knowing it and thereby perhaps as not in a position to endorse the major premise in the husband’s alluded to Practical Inference, which in turn provides Steering Thrust against Delaying the Deposit, which would be acting on such presumed-knowledge. (The more precise characterization is Posting an approval for such a preferred action, rather than signaling; see my (2015e)).
He thereby also steers her towards their now common preferred action. It appears not to be necessary, but he modulate the strength of such steering of his as is called for and he deems fit , although largely dispensable practically, which seems minimal.
‘Signaling’ or ‘indicating’ are not part of the conveyed content. Rather, they invoke inferences that hearers can be expected to gauge, or even make, at such points as here, on the basis of the conveyed contents in the context. The precise terminology that is requisite here is Posting -- the Pragmatic Stances that the speaker posts himself as having in view of how he steers, which are public and thus noticeable. For a more detailed account of Posting, see my (2021a).
Skad-variants are classified by their key term: A ‘sure’ variant (viz.: Now I am not sure that p), or a ‘might’ variant (viz.: It might be that –p), or as embedded in ‘I believe’ , etc. More precisely, the Skad as introduced above is the knowledge-variant.
First, more precisely, the speaker steers away from an action and Posts himself as not approving such an action. Given that he is taken to be candid, the hearer would be in a position to conclude that he wouldn’t take that action, thereby posting himself as what can be taken as retracting; see my (2015e). Second, in this paper, as noted, I focus on a two-stage presentation of the puzzle, as in the original version of the bank case. But the same, mutatis mutandis, holds if the puzzle is framed in terms of distinct cases, corresponding to the first and second stages respectively. In this framing there is no retraction, but the Pragmatic function of the Skad in such a presentation is along the same lines: To steer towards the requisite action and to so Post himself as not endorsing p as a premise in the alluded to Practical Inference, and accordingly as not sayable for them (due to its would-be Steering Thrust in that context), thereby undermining the practical reasoning via the practical inference alluded to.
More precisely: So signaling (or better: Posting) that the hearer is in a position to conclude that the speaker now has a new preferred-action, although the speaker may have now moved to an undecided position with no preferred action . In cases where there are practically only two actions on the table (which are incompatible), retracting one suggests or indicates an adherence to the other. (Cases of this sort might also be ones where there is a ‘degenerate’ third option of Do Nothing, which either coincides with one of the actions on the table or else might be inferior to each of the two other actions, and hence ignored.).
And similarly by signing on to it, which engenders a weaker Steering-Thrust than endorsing (as does 'settling on' it). Saying and endorsing a major premise steer towards acting on it (but the ‘it’ needs to be sufficiently specific, which is often not so, in particular without further elaboration on the alluded-to Practical Inference, and doesn't cover various Pragmatic Stances other than AD). ‘act[ing] on p’ is different than the steering-thrust produced by saying it: acting is what a speaker does, whereas Steering Thrust is transmitted towards a hearer, as part of the conveyed content. They of course would usually coincide. The steering thrust is also modulated by a degree, unlike ‘act on p’ (which is a fixed, unmodulated idiom), and the action in question is fixed in the Steering Thrust (is a component of it—but not sufficiently by the explicit content of ‘act on p’), as noted above. The AD-content transmitted by a certain verbal act is steering thrust towards an action, and thus not propositional and devoid of truth-value, but with a certain degree of strength, whereas acting on p might be instantiated by such a verbal act, or not, or only partly. ‘acting-on-p’ is used in AD-Pragmatic contexts but need not in other Pragmatic domains of Steering-Thrust Pragmatics which don’t involve actions (e.g., Informational, Attitudinal, Emotive Pragmatics), whereas Steering Thrust is the common denominator in them all and results from such injunctions. Hence the limitations and merely local applicability of ‘act on p’, and its being locally instantiated by Practical Inferences of certain types.
Such as 'I am now not sure that I know' or 'It might be that I don't know', etc.
Not just in Action-Directed Pragmatics, but also in other Pragmatic domains. But the degree of the epistemic position committed to is often exaggerated when the Pragmatic function is operative as well.
Yet subject to further additional modulation of the Steering Thrust conveyed, which can be done linguistically, by adding a refinement such as ‘very well’ (yielding ‘know very well’), or, primarily in verbal interactions, by non-linguistic modulators such as intonation, gestures and facial expressions. The Steering Thrust conveyed can yield significant pressure, epistemic and/or non-epistemic.
When it serves this double duty, the seemingly presented epistemic position is prone to exaggeration (and thus is only roughly indicated and is not fully committal); for a more detailed presentation, see my (2020a).
As noted, Pragmatic signaling by the speaker to the effect that he is not in favor of a certain action needn’t commit him as to how or why he reversed his preferred action (in case he did), though conversational participants can easily figure it out, in such a case. Such decision process is not reflected in the Pragmatics.
If at all; he could just say: Ok, so I am ready to Deposit Now. Accordingly, there is no need to use Knowledge-Ascriptions in generating the puzzle, though one can and thereby pose the seeming inconsistency between self-Knowledge-Ascriptions. The seeming inconsistency could be generated by different seemingly inconsistent sayings, e.g.: ‘The bank is open on Saturday’ (first Stage, or by prefixing the operator ‘I can tell that...) vs. ‘Now I can’t tell that the bank is open on Saturday.’ This way knowledge wouldn’t be an explicit source of the seeming inconsistency (unless one invokes a Knowledge Norm of Assertion). Note too that using a knowledge ascription in the first stage by either the husband or wife might be less than natural or quite unnatural: It’s natural for the husband to say (in Stage I): I am sure the bank is open on Saturday. It’s not at all natural for the wife to ask him ‘Do you know that the bank is open on Saturday?’ -- it would be more natural for her to ask ‘How do you know that?’. However, his answer is highly unlikely to change her action-preference, and she can confidently anticipate that once she informs him about the mortgage, he’ll change his current preferred-action and converge to hers. (Her response, if not just unnecessarily argumentative, would make more sense and be more natural if she were to attempt to delay informing the husband about the mortgage, or entirely hide it from him.)
This treatment applies fully, mutatis mutandis, to the other variants of the same type, such as Fantl and McGrath’s local/express train or Cohen’s airport cases.
Our concern has been to deal with the various puzzling aspects of such cases, which, I have claimed, are primarily Pragmatic—neither semantic nor decision-theoretical, and can be satisfactorily resolved just with an adequate Steering Thrust Pragmatics, and in particular its AD branch.
Very roughly, normative pressures may allow for a ‘relaxation’ and a certain extension of epistemic thresholds, as here. This is a local feature of the large phenomenon of the functioning of the Overall Sayability function; see above, section 5, around 4th paragraph. That is, even if he didn't know it but was in an epistemic position close to it, the Pragmatic context accommodates such a use, since, even though it's exaggerated epistemically, it serves the right function Pragmatically.
In simple Common Sense.
I emphasize again that I don’t attempt to account here for the husband’s deliberation that has led to the reversal of his preferred action. It’s not a case of a simple Practical-Inference, and there are biases involved. For my approach regarding this issue, see my (2015c) and (2015d). (In (2015d) I explain what a ‘simple’ Practical Inference is, as part of a classification of Practical Inferences of varying complexity in the Common Sense as well as outside the Common Sense.)
However, again, the Pragmatic treatment offered here doesn’t attempt to treat such a decision-reversal—only to explain what the husband says as a result of it: Which Steering Thrust he conveys in the service of this new preferred action, which amounts to the implicit content he conveys that is dominant here.
If the major premise becomes: ‘There is high risk associated with the Delay-Deposit action’, this is no longer a case of simple Practical-Inference (in my classification), and requires a shift to a different heuristic—a move of what I call an inferential ascent; but I can’t dwell here on the husband's deliberation leading to his new Preferred Action, as mentioned, nor on such differences and moves; see my (2015d). However, it seems, one can consider such a case as a case of acting on ‘There is a high-risk if the bank is not open on Saturday’.
Or, better, that he no longer knows it, since I focus on the 2-stage version of the bank-case.
Note that I consider the bank-case as in practice a two-action type case (viz., DN and DD), treating the option of suspending judgment about what to do as not viable for a competent speaker.
Nor, for that matter, just epistemically warranted Sayability. Recall that in mixed-norms contexts what governs sayability (and thus sayability intuitions) is the Overall Sayability function, which gives, say, varying weights to the various normative components (in contexts of this sort).
Even if we construe the case, as is my natural construal, so that the husband knows (or at least, uncontroversially, knew) that the bank is open on Saturday (at least in stage I). However, this construal of the puzzle vis-a-vis his epistemic position then and there (reflecting my steady Pragmatic approach in opposition to Fantl & McGrath’s Shifty-Epistemology construal) is not at all crucial for the puzzle and its resolution along the lines pursued here: We may leave it undetermined, between knowledge and falling evidentially somewhat short of what's required for knowledge—but with sufficient evidence to warrant his epistemic Sayability of only ‘The bank is open on Saturday’, or have evidence somewhat weaker than what’s required for it. This way, there is no need to invoke here an Epistemic Norm of Knowledge. Furthermore, the puzzle remains regardless of whether we assume that the content clause itself is not true: The puzzle stems from the incompatible, yet intuitive, use of saying that p first and then the seeming retraction to saying ‘I now don’t know that p’ later, without any evidence pertinent to it in between. The mistake is in construing it as an epistemic retraction rather than as an AD-retraction—a Pragmatic retraction.
Willy-nilly, as by Stanley. This is the common mistake of the various strands of Pragmatic Encroachment. DeRose also considers WAMs, in response to objections (but not at the beginning) and dismisses them, for the wrong reasons (in my opinion, as I elaborated in my (2012e): “On DeRose: The Worry about WAMs given Action-Directed Pragmatics”, and in the literature; see also note 50 above).
Or as tracking Epistemic Sayability.
In a particular way, with the instrumental norms, now with a much higher intensity, suppressing specific semantic/epistemic normative pressures. Note that normative pressures of different sorts enter the normative profile of the case. The mistake in question lies in the Pragmatic-Encroachment construal of the husband’s denial of the knowledge-ascription in Stage II as ‘intuitively true’ and thus true, which is a semantic construal, conflicting with the knowledge-ascription utterance in Stage I (of: I know that the bank is open on Saturday), construed as true as well. But beyond the pros and cons of Pragmatic-Encroachment and my Pragmatic approach, which is our central concern in this paper and comes to a head in Stage II of the puzzle, one can maintain that, as I noted above, that utterance is a borderline case—that it’s undetermined whether it’s true if the pertinent epistemic-position the husband had (construed traditionally) is less than what’s required for knowledge. The puzzle arises whether or not the husband is in an epistemic position to say that he knows in that situation (in either epistemic version—of sufficient evidence for knowledge or of his evidence falling somewhat short of it).
A difficulty for Pragmatic Encroachment arises if we modify the case so that we, the audience, but not the conversational participants in the vignette, know that the bank is closed on Saturday (contrary to the husband's evidence). But now, the intuition that what the husband says in Stage I is true dissipates, and with it presumably, on the Pragmatic-Encroachment approach, also our inconsistency. But intuitively, the puzzle remains, since so does the seeming inconsistency (of the husband). On my Pragmatic approach, the treatment is just the same: In what the husband says in Stage II he conveys only Steering Thrust, not a proposition, and Steering Thrusts are devoid of truth values. Thus, there isn’t even the appearance of a semantic inconsistency. Rather, a sense of inconsistency is generated by the conflicting Steering Thrusts in the two stages—and this aspect remains intact as in the both versions, regardless of the respective truth-values. The inconsistency is only a seeming inconsistency since the husband received in between information relevant evidentially for his being in a position to convey the Steering Thrust he did in the second stage—the information about the risk. Consequently, he appropriately reversed his preferred actions and accordingly his Steering Thrust, which is action-directed. This therefore explains the sense of inconsistency, and also why it’s merely apparent.
The expressed content is the semantic/literal content. Of course, philosophers have been aware of Pragmatic inconsistency, most notably Moorean sentences/utterances, primarily in cases with an indicative utterance a prefix of a denial of some epistemic operator is added, such as ‘believe’ or ‘know’. But the literature on this subject has ignored the major steering character of the inconsistency—that by saying that p, in AD contexts (which are our concern here, though the Pragmatic phenomenon is general), I steer my audience towards acting one way whereas if I add right away ‘but I don’t believe that p’, then, apart from any semantic or epistemic inconsistency, I thereby steer away from it (e.g., towards not acting that way). By saying that p I s steer (directly) towards an AD position of ‘settling’ on the p-contingency, but by saying the second conjunct I undermine my settling on ‘p’. In an AD-case, the (composed) conveyed Steering Thrust (by saying the first conjunct) is towards performing the action in the ‘impending’ simple Practical Inference with p as a premise (on which I presumably ‘settled’, epistemically). By the second conjunct, since I thereby undermine my settling on the p-contingency, I also undermine my ability to run through that Practical Inference, and therefore undermine my (indirect) Steering Thrust to act on p; and similarly in the ‘know’ variant. But note that, even though in this paper we focus on AD-Pragmatics, Moorean inconsistencies hold generally (as in Informational Pragmatics, AD- Pragmatics, and Ethical Pragmatics—see, e.g., my (2013), (2018c), Section 11, and (2020b). So apart from semantic inconsistency concerns, such Moorean sentences constitute cases of Pragmatic inconsistencies—in the sense of Steering Thrust Pragmatics -- of conflicting Steering Thrusts. The paradigm of only Pragmatic Inconsistency (without semantic inconsistency) is Stanley’s ‘certain’ case, viz., of saying “p, but I am not certain that p”, exemplifying the hallmark of Pragmatic Inconsistencies—opposing Steering Thrusts (e.g., for performing an action and away from it, but more generally as well). (For a development of this point, see my “Pragmatic Inconsistencies” (2013). For an extension to Informational Pragmatics, see my (2018c), Section 11. For the extension of Steering-Thrust Pragmatics to Ethics, see my (2020b). The main point there is that ethical statements in the object-language convey ethical Steering-Thrust towards Ethical positions or stances, which constitutes their Pragmatic conveyed content and is their main content; consequently, they are devoid of Truth-Values.).
E.g., if he conveys a message with a dominant AD-content (or another non-epistemic/semantic type of dominant content) but whose semantic content is suppressed—not conveyed, after using an utterance whose semantic content (in the context) is indeed semantically incompatible with that of the other saying under consideration.
Thus, he need not be inconsistent simpliciter in case the two sayings (including assertions) have been made at times that are significantly apart from each other, and he has acquired an adequate rationale for reversing or considerably changing his approval or disapproval for being in the targeted Pragmatic position (of his steering) of the first saying, e.g., in AD-cases, regarding performing his Preferred Action, which has changed. Note that such messages can, but need not be, verbal, fully or partly—though my focus here is on Linguistic Pragmatics (that requires that the implicit messages under consideration be conveyed by linguistic verbal-acts—and as such with linguistic/semantic explicit content). Of course, we have long been aware of Moorean utterances whose semantic contents are consistent, yet the utterances are not, rendering the speaker as inconsistent—viz., Pragmatically inconsistent. On my account, Pragmatic inconsistencies are thereby generated falling in the domain of Steering Thrust Pragmatics -- with the conflicting Steering Thrusts (which are the Pragmatic contents) via the two conjuncts, even though the literal contents are indeed conveyed by the conjuncts (and thus not suppressed). Yet a change in the normative context can change the Pragmatic content conveyed by (at least) one of the conjuncts (said temporally separately). For elaborations, see my (2013). (I don’t deal here with the important subject of inconsistency (or incongruence) due to the tension created by incompatible Postings since this paper doesn't deal with the important Pragmatic phenomenon of Posting—see my (2015e).)
Where the very familiar Gricean implicatures are absent. I have argued that attempts to show that Gricean implicatures are the culprits (of such puzzles) have not been successful (see, e.g., Pritchard (2005), Brown (2006) and my (2012a) and (2015a)). The kind of AD implicit Pragmatic-content that is our main concern here isn’t, I take it, part of what Gricean Pragmatics has focused on (and the same for Bach, e.g. (2008), and Stalnaker, e.g. (1999)).
This of course is a form of Stanley’s certainty Moorean-sentence (Stanley, 2008). Under my Pragmatic construal, if the context were with considerable operative Pragmatic pressures (e.g., AD), the husband in such a case could be construed as Pragmatically inconsistent, or else, as conveying one Steering Thrust first (by the first conjunct), but retracting it later—fully or partly, as in a dynamic context (due to a change of mind). But I have just stipulated (in the text) that the context is primarily epistemic, due to quite low instrumental pressures. Incidentally, in view of the husband's being consistent in uttering his conjunction to his friend (privately), this is an argument why being certain is not an epistemic norm of assertion, contra Stanley. Although I haven't dwelt on it in this paper, in epistemic contexts too there are usually Pragmatic contents that are conveyed—as Steering Thrusts towards epistemic positions. (This is what is more precisely meant by semantic/epistemic contents being conveyed—relatively purely, as in such 'pristine' context: What a speaker conveys (to someone), even in such context, is a Pragmatic phenomenon.) Pragmatic inconsistencies need to be recognized as distinct from semantic or epistemic inconsistencies; see my (2013).
For counter-examples to Epistemic Contextualism which employ such variations, see my (2012c). Note that 'He knew that p although he wasn't entirely sure' is semantically consistent.
And he may very well have not violated the Sincerity Norm (see my (2021a) and (2021b)). For more on such types of leeway and how they are accorded, and in particular in conjunction with the role of Pragmatic functors such as the know-Skad, e.g., ‘I don’t know that the bank is open on Saturday’, presented above, see my (2015f). The function of the operator ‘I know’ in the husband’s response to his wife is, on this variation, primarily Pragmatic—it serves to convey a stronger Steering Thrust (towards Delaying the Deposit) than the mere indicative form without that operator, which conveys a weaker AD-Steering-Thrust.
The Pragmatic puzzle invoked here, in my view, can be a composite—composed of several components that can but need not overlap, and require separate treatments. Apart from the Pragmatic perspective emphasized and presented in this paper, one should be aware of the following three potential components as well: First, a shortcut function of Skad-like locutions, where, e.g., ‘I now don’t know that p’ functions as a shortcut for ‘I now can’t tell that p’ or ‘I now can’t tell that I know that p’. This construal garners further plausibility once we notice that a literal Skad-like (such as in the husband’s retraction, if it were taken to also convey its literal content, contrary to my position) might very well be epistemically unwarranted (regardless of the Pragmatics aspect), since a low or even a moderate decline in epistemic position is not prone to warrant a full epistemic retraction from ‘I know that p’ to ‘It’s not the case that I know that p’, which is its literal negation. This is so since one can epistemically retract only to ‘Now I can’t tell you whether I know that p’, i.e., to a suspension of judgment regarding whether one knows that p (of course, on the non-Pragmatic-Encroachment construal of 'know'). There was no change in evidentially relevant information in our case, and therefore a retraction, let alone a full retraction, of the original statement is not epistemically warranted. (Succumbing to a bias, although prevalent, is still not epistemically warranted, even if excusable;. And a considerable loss of confidence is not warranted cognitively (in terms of cognitive competence, of which epistemic competence is a component) except perhaps for the first couple of times of experiences of that sort. More generally, a low or moderate decline of evidence regarding the original statement can leave it open epistemically whether or not one can still tell that one knows that p, but is not prone to warrant epistemically a full reversal of the self-knowledge-ascription. In usual formulations of the puzzle (especially DeRose's later formulation) there is a mixture of a Pragmatic retraction and an epistemic retraction. Yet these are two distinct types of retractions, although they can and often do overlap. But the puzzle, I claim, is due to the contribution of the Pragmatic retraction, which needs to be isolated in order to be brought out in a precise form. The original puzzling element remains in tact in the isolated Pragmatic retraction version, but dissipates in an isolated version of the pure epistemic retraction that also underpins the mixed case. For Pragmatic Retraction vs. Epistemic Retraction, see my (2015d).
I don’t consider a loss of confidence an epistemic failure, but rather a cognitive defect, which may be excusable—i.e., a result of a species-wide cognitive shortcoming. This would explain and may cognitively excuse a retraction of a self-knowledge-ascription (for ‘p’) if, e.g., the speaker’s confidence in his evidence (for p) has also declined so that accumulatively they warrant a confidence level only somewhat below the knowledge threshold (for ‘p’) but not a much lower level—e.g., not the lower level of mere warranted-belief. Then he can remain warranted to straddle the self-knowledge-ascription without committing neither to it nor to its negation, i.e., to suspend Sayability about each. On my (non-Pragmatic-Encroachment) position, in a usual bank-type puzzle case, the husband continues to know that the bank is open on Saturday all along, and he is not epistemically warranted to commit to the literal negation of what he has said first even if, due to confidence loss, he can’t tell now that he knows. And since his response is ‘intuitive’ and this response is fairly standard, there might well be a construal of the literal self-knowledge-denial on which he isn’t committed to its literal construal despite saying it, thereby avoiding an error theory about such cases. And indeed, there seems to be a widespread use of this sort. This option fits the Pragmatic position I develop here (see also the last section). But note that I focus here on versions of bank-type cases that exhibit just Pragmatic retractions, which nevertheless can accommodate mixed uses of first-time cases of mild (or even often moderate) cognitive confidence-loss (where the subject doesn't lose his epistemic Sayability that p, and of course not his belief that p, despite a modicum of a cognitive retraction—i.e., epistemically unwarranted mild cognitive retractions that are cognitively excusable) so long as the explicit content of the Skad is construed literally (and not as a shortcut).
But of course, many cases are and can be mixed cases of Pragmatic retraction, epistemic retraction, and cognitive retraction, with various degrees. Note first that, where there is only minor confidence loss in not-merely epistemic contexts, ‘I am not in a position to tell’ may very well invoke non-epistemic norms, such as instrumental or AD norms, and thus need not reflect only operative epistemic norms. Thus, it isn’t a purely epistemic modal, and can be responsive to various normative pressures such as, in our case, AD norms. (Note that the locution 'I can't tell that p' rings as an epistemic modal to my ears—unlike here on a 'I am not in a position to tell that p'.)
Second, to expand on a previous remark here, the subject may well experience an epistemic/cognitive Confidence Loss once he hears about the risk, which is prone to take place in the first (or few) couple of encounters of that sort that , resulting in a surprise, a momentary confusion and/or a need for integration, reflection, and ‘pulling oneself together’ cognitively (compare experiences of Hard-Core Skeptical encounters). Once such a confidence loss has been encountered in a few cases, the subject would not be prone to further significant-enough Confidence-Loss experiences in similar cases (or increasingly less so).
Third, I have argued that a prominent, common and entrenched cognitive bias underpins much of the first-time responses to such high-risk and surprise-invoking cases, which I called ‘The Risk Bias’. This bias is primarily to the effect that, when news of a considerable risk reaches the subject, then (in general, and especially if she is not very sophisticated) an ingrained, primitive cognitive way of coping with the high risk drives the subject to shift her assigned likelihoods: To drive up the likelihood of the risky course (viz., in our case, that the bank is closed on Saturday) and accordingly to shift down the likelihood of the complementary course (viz., that it’s open). But such a change of likelihoods is epistemically and probabilistically unwarranted (and is prone also to distort other related likelihoods); yet it’s an adaptive bias (from the perspective of ‘cognitive evolution’).
I elaborate on these various options and mechanisms (including the distinction of (purely) Pragmatic retractions vs. Epistemic retractions) elsewhere, primarily in my (2015b) “Abstract: Cognitive Risk-Bias and the Threat to the Semantics of Knowledge-Ascriptions”, (2015c) “Cognitive Risk-Bias, Confidence Loss, and Knowledge-Ascriptions” and (2015d) (on Pragmatic vs. Epistemic Retractions).
As noted, I allow in the bank cases a variant where the subject is aware (or at least merely subjectively aware) that he knows that p as well as the option where he does not know nor take himself to know that p but is in a position to assert that he knows due to the ‘leeway’ accorded in Pragmatic contexts. So the husband’s evidence spelled out in the example may well be sufficient for Knowledge (and/or for the subject to take himself as knowing—here I am straddling the distinction between knowledge and subjective knowledge). But even if it's not—it could be slightly below (not sufficient for knowledge/subjective-knowledge) but with 'I know that p' being sayable nonetheless due to the extra AD-pressure and the associated ‘leeway’. The surface issue about the seeming inconsistency arises in both variations, but it's clearer in the version where he does know (or at least has had subjective awareness that he knows—in stage I): If he doesn't, there is still the seeming inconsistency about his two sayings (on the literal—semantic—reading of what he says in stage II), which in such a case can be put to rest by bringing out that in fact he is not in an epistemic position to say that p. So it's only what he says that raises the seeming inconsistency - but not what he is epistemically in a position to say. But of course this treatment is dwarfed by the explicit content not being conveyed in stage II. But it’s the second stage that provides the real puzzle. Is he inconsistent epistemically? What he says in stage II, read literally, commits him to subjectively having known previously (by the word ‘now’). So on the literal version he may be misleading (via the literal content) or seems insincere, and he seems to be inconsistent -- which of course intuitively he isn't. If he doesn't know that p but isn't aware of it and thinks that he does, we are back to the inconsistency issue as in the case where he does know: The inconsistency puzzle is independent of whether he in fact does or does not know. Consequently, the puzzle is not just about knowledge—it has a parallel Sayability version. But Shifty Epistemology doesn't handle this deeper puzzle—it only shifts the semantics of ‘know’. Yet for the inconsistency (or other impropriety) about Sayability here there is no semantic key for resolving the issue (given no knowledge requirement for mere sayings, which need not be assertions)!
Even though sayable simpliciter. It’s not epistemically sayable by him in stage II since we assume that in stage I he has known it and/or-at-least was then in an epistemic position to tell that he does. (However, such an awareness may have dimmed and (even duly) suppressed in such cases—pushed to a 'cognitive back-stage'). In view of our Pragmatic perspective, the explicit content of what he has said in stage II is suppressed and not conveyed Pragmatically. Note that if we assume (contrary to the standard version as well as to our version here) that he not only has known it, outright (not just being merely close to knowing), but also that he has been in a readily accessible epistemic position to clearly tell that he knows it, then he is not in a position to say that he doesn't know,—this is an epistemic constraint he is committed to (which gives rise to an 'extended leeway' -- that he doesn't know too well). But otherwise, if he merely knows it or is close to knowing it (epistemically -- say, with the sort of evidence usually assumed), as in the usual version is that we have been working with, then, although what he has said in both stages makes him seem inconsistent, he need not be if in the first stage he misleads, misrepresents himself and perhaps lies. In this case, he need not be epistemically inconsistent, though nevertheless what he has in both cases is inconsistent (if read literally). But of course, whereas his epistemic consistency (apart from 'what he has said') in the standard version is the puzzle, it's not in the above version where he cheats, since then the content he conveys (when informative, as in stage I -- steering towards an epistemic stance,) doesn't represent his own epistemic stance, since he is insincere. In this case, he is not in an epistemic position to say what he does in the first stage given that he is aware that he doesn’t know it (and isn't in an epistemic position to say even the content clause). Now back to our usual puzzle case.
Note that the husband also need not be inconsistent if he just has not been able to tell whether or not he knows it. In such a case, he has not been in an epistemic position to say that he knows—it’s not epistemically sayable by him, although he does (in stage I); and in such a case, the puzzle still arises: How come it’s intuitive for him to nonetheless say what he does in either stage? If he is not in an epistemic position to say whether he knows, he is also neither in an epistemic position to say that he does (as he does in stage I) nor that he doesn't know (as he does in stage II). But we have also resolved the stage I case -- the allowed 'leeway'. In stage II, it's not epistemically sayable for him to say what he does (since he doesn't know whereas he can't tell whether he does), but in this case, unlike our standard case, it need not be because (or just because) he doesn't convey the explicit content of what he does: Here, the shortcut construal seems more plausible (which in turn increases its plausibility in general); see my (2015b) and (2015c). This is one way of bringing out a semantic construal of a particular sub-case, which supplements its Pragmatic profile—it doesn't compete with it, but more importantly, it doesn't generalize—unlike our Pragmatic treatment. (As noted, the case can be different if you assume considerable Confidence Loss, as usually in the first couple of encounters of this sort. But the puzzle still arises, even then, first since subjects are appropriately prone to say what they say in the puzzle even when they have already overcome the hesitation of the initial instance or initial instances due to the Confidence Loss: Confidence Loss limits somewhat the applicability of the puzzle, but doesn’t undermine it. But second, since Confidence Loss is like in our above case where the subject can't tell whether he knows, and our above treatment applies. This however is a more complex case since in it, unlike the other case, the subject is affected by a Cognitive Bias, which allows him epistemically to say what he does only excusably, though not strictly speaking.)
To underscore what we have seen in the case where the husband is not quite in an epistemic position to tell whether he does or doesn’t know it: he might nevertheless be in a position to say what does (as in Stage I): The implicit (Pragmatic—AD Steering Thrust) content is the same (whether he does or doesn't know or can't tell), but it isn’t the entire, overall, conveyed content, since in such a case the non-epistemic (instrumental -- AD) norm has a sufficient weight: In such a case, where the husband can't tell that he knows (by way of just falling short of being able to tell that he does), there is AD Steering Thrust conveyed by what he says in the first stage. But the implicit content he conveys doesn’t overweigh the (epistemic) explicit content (which is indeed conveyed) but is aligned with it, and the husband is not fully committed to the latter (due to the 'leeway'— he is allowed to exaggerate, while posting a fuzzy epistemic commitment).
Note again that a loss of confidence may render him not in an epistemic/cognitive position to say that he knows, but this doesn’t equip him with epistemic sayability that he doesn’t know it: At best, he may be in a cognitive (and thus Overall) position to say that now he can’t tell whether or not he knows it, since his loss of such epistemic Sayability is also due to a ‘cognitive malfunction’, not a pure epistemic mistake, which is prone to be temporary (he will after a while regain a better sense of his epistemic position, which will approximate that of a subject who hasn't suffered Confidence Loss). which, However, unless the Confidence Loss is very massive (which would be unusual), he still keeps believing that p (viz., that it’s open) even though it feels to him that he doesn't know it.
I.e., with other operative norms being dominant; and also in various Pragmatic contexts with dominant non-epistemic norms, more generally.
In this paper I use ‘context’, whereas the more precise term that I prefer is ‘setext’ (i.e., conversational context together with physical setup); but I don’t have the space here to elaborate. I introduce it in my (2018b) and (2018c).
However, it is nevertheless subject to certain (relatively loose) epistemic constraints with an extended epistemic leeway, specified elsewhere (see my (2015f)): The extended epistemic leeway is granted to the Skad so that the appropriate Pragmatic message can be conveyed: Without that epistemic leeway, its use could count as distorting the case, specifically regarding how the subject posts his pertinent epistemic position. An extended leeway regarding how the subject posts his pertinent epistemic position is designed to not render how he posts himself as incompatible with the literal content of his Skad: Even though the explicit content is not conveyed by the Skad, it does have such an effect on how the speaker posts his respective epistemic position. Accordingly, the ‘upper-bound’ of his pertinent epistemic commitment is adjusted upwards: The subject can then be in a position to post himself as having (and thereby being committed to) a much looser upper bound of his pertinent epistemic position (in his use of the Skad): The upper-bound of his posted pertinent epistemic position is shifted upwards to that of not knowing-very-well (which is way less restrictive than just not knowing).
I argue for this point in detail in my (2015f), and I don’t have in this paper the scope to elaborate on the full resolution of the Inconsistency puzzle for us—see Sect. 1 above—as opposed to the husband’s alleged inconsistency; see the version of the puzzle in footnote 9; see also footnotes 1, 23, and 35. For the Pragmatic phenomena brought out by that version of the puzzle, see my “The Audience Norm of Saying” (2012d).
Or proposition-like relata. Yet implicit Steering Thrust can be incompatible, or incongruent, with another Steering Thrust, if, e.g., it differs from the latter only in polarity (and can be also incongruent with an explicit content—e.g., in Moorean utterances.). And indeed, the Steering Thrusts the husband conveys in stages I and the II have different polarities. Yet the husband is still Steering-Thrust-congruent (or consistent), since his changing the polarity of his Steering Thrust is a result of a change of his Preferred Action, which is rational: New information about high risk might very well appropriately affect the need for an action other than the action previously steered towards.
For a full discussion of Pragmatic Inconsistencies, see my (2013).
Again, here can I assume that he knows the bank is open on Saturday (though I leave it open in other places). But the example’s bite remains intact if we assume that the epistemic threshold of saying is lesser than that of assertion (which I consider right, whether or not the Knowledge Norm of Assertion holds) and the subject is thus epistemically entitled to say that the bank is open on Saturday even though he falls short of knowing it as well as of being in an epistemic position to assert it.
At stage I, both the explicit content and the implicit content are conveyed.
References
Bauman, P. (2011). Wams—Why worry? Philosophical Papers, 40, 155–177
Bauman, P. (2012). Knowledge, practical reasoning and action. Logos & Episteme, 3, 7–26
Bach, K. (2008). Applying pragmatics to epistemology. Philosophical Issues, 18, 68–88
Blome-Tillmann, M. (2020). What shifts epistemic standards? DeRose on contextualism, safety, and sensitivity. Episteme, 10(1), 21–27
Blome-Tillmann, M. (forthcoming). Subject-Sensitive Invariantism, High-Stakes/Low-Stakes cases, and Presupposition Suspension. Episteme.
Brown, J. (2005). Comparing contextualism and invariantism on the correctness of contextualist intuitions. Grazer Philosophische Studien, 69, 71–99
Brown, J. (2006). Contextualism and warranted assertibility maneuvers. Philosophical Studies, 130(3), 407
Buckwalter, W. (2010). Knowledge isn’t closed on Saturday: A study in ordinary language. Review of Philosophy and Psychology, 1(3), 395–406
Cohen, S. (1987). Knowledge, context, and social standards. Synthese, 73, 3–26
Cohen, S. (2004). Knowledge, assertion, and practical reasoning. Philosophical Issues, 14, 482–491
DeRose, K. (1992). Contextualism and knowledge attributions. Philosophy and Phenomenological Research, 52, 913–929
DeRose, K. (2002). Assertion, Knowledge, and Context. Philosophical Review, 111, 126–203
DeRose, K. (2009). The Case for Contextualism: Knowledge, Skepticism, and Context. Oxford University Press.
Dinges, A. (2016). Epistemic invariantism and contextualist intuitions. Episteme, 13(2), 219–232
Engel, P. (2009). Pragmatic encroachment and epistemic value. In A. Haddock, A. Millar, & D. Pritchard (Eds.), Epistemic value. (pp. 183–203). Oxford University Press.
Fantl, J., & McGrath, M. (2002). Evidence, pragmatics, and justification. Philosophical Review, 111(1), 67–94
Fantl, J., & McGrath, M. (2009). Knowledge in an uncertain world. Oxford UP.
Fantl, J., & McGrath, M. (2012a). Replies to Cohen, Neta and Reed. Philosophy and Phenomenological Research, 85(2), 473–490
Fantl, J., & McGrath, M. (2012b). Pragmatic encroachment: It’s not just about knowledge. Episteme, 9(1), 27–42
Fantl, J., & McGrath, M. (2012c). Contextualism and subject-sensitivity. Philosophy and Phenomenological Research, 84(3), 693–702
Fantl, J., & McGrath, M. (2012d). Arguing for shifty epistemology. In J. Brown & M. Gerken (Eds.), Knowledge-ascriptions. (pp. 55–74). Oxford: Oxford University Press.
Gerken, M. (2017). On Folk Epistemology. Oxford.
Goldberg, S. (2015). Assertion: On the philosophical significance of assertoric speech. Oxford University Press.
Goldberg, S.(forthcoming). Stakes, practical adequacy, and the epistemic significance of double-checking. Oxford Studies in Epistemology.
Grice, H. P. (1989). Studies in the way of words. Harvard University Press.
Hawthorne, J., & Stanley, J. (2008). Knowledge and action. Journal of Philosophy, 105, 571
Kahneman, D. (2011). Thinking. Penguin Books.
Kim, B. (2016). Defense of subject-sensitive invariantism. Episteme, 13(2), 233–251
Kvart, I. (1993). Mediated reference and proper names. Mind, 102, 611–628.
Kvart, I. (2006). A probabilistic theory of knowledge. Philosophy and Phenomenological Research, 72, 1–43
Kvart, I. (2008) The multifariousness of the internal-external distinction (manuscript 1).
Kvart, I. (2012a). Rational assertibility, the steering role of knowledge, and pragmatic encroachment. PhilPapers, pp. 1–51.
Kvart, I. (2012b). The non-gradability of 'know' is not a viable argument against contextualism. PhilPapers, October, 2012, pp. 1–7.
Kvart, I. (2012c). A counter-example to SSI and contextualism. PhilPapers, posted October 2012.
Kvart, I. (2012d). Audience norm of saying. (manuscript-2).
Kvart, I. (2012e). On DeRose: The worry about WAMs given action-directed pragmatics (manuscript).
Kvart, I. (2013). Pragmatic inconsistencies (manuscript-3).
Kvart, I. (2015a). Action-directed pragmatics secures semantically autonomous knowledge. June 19, PhilPapers, posted August 2014.
Kvart, I. (2015b). Abstract: Cognitive risk bias and the threat to the semantics of knowledge-ascriptions. PhilPapers, posted June 26, 2015.
Kvart, I. (2015c). Cognitive risk bias, confidence loss, and knowledge-ascriptions (manuscript-4)
Kvart, I. (2015d) Pragmatic vs. epistemic retraction, common-sense decision heuristic of practical inference, and inferential ascent (manuscript-5)
Kvart, I. (2015e) Posting and Presupposing (manuscript-6).
Kvart, I. (2015f). Steering-thrust pragmatics, shifted leeways and pragmatic functors with negative polarity (manuscript 7).
Kvart, I. (2017) Real vs. pseudo pragmatic encroachment and the ‘quick jump’ fallacy. (manuscript-8).
Kvart, I. (2018a). A high token-indicativity account of knowledge. Acta Analytica, 33(3), 385–393.
Kvart, I. (2018b) Formal pragmatics (for action-directed contexts) as an coding conception. PhilPapers.
Kvart, I. (2018c). Informal motivation for Action-Directed Pragmatics. special issue of Philosophical Perspectives—Philosophy of Language. John Hawthorne and John Turner (eds.), 32, 219–253.
Kvart, I. (2018d) Elaborated formal pragmatics (for action-directed contexts) (manuscript-9).
Kvart, I. (2020a). The steering thrust phenomenon in action-directed-pragmatics. In R. Borges & E. Sosa (Eds.), Synthese special issue: Knowledge and justification: new perspectives.
Kvart, I. (2020b). Ethical pronouncements belong to steering-thrust pragmatics (manuscript 10).
Kvart, I. (2020c). Do knowledge and stakes flip in tandem? Vicarious audience: A key mistake in ascribing not-knowing in action-directed situations (manuscript-11).
Kvart, I. (2020d). McGrath on knowledge attribution and pragmatic encroachment (manuscript-12).
Kvart, I. (2021a). Posting and presupposing (manuscript 13).
Kvart, I. (2021b). Assertions and norms of assertion (manuscript 14).
McGrath, M. (2007). On pragmatic encroachment in epistemology. Philosophy and Phenomenological Research, 75(3), 558–589
McGrath, M. (2018). Defeating pragmatic encroachment? Synthese, 195(7), 3051
Nagel, J. (2008). Knowledge-ascriptions and the psychological consequences of changing stakes. Australasian Journal of Philosophy, 86, 279–294
Pritchard, D. (2005). Contextualism, scepticism and warranted assertibility manoeuvres. In J. Keim-Campbell, M. O’Rourke, & H. Silverstein (Eds.), Knowledge and skepticism. Cambridge: MIT Press.
Reed, B. (2010). A defense of stable invariantism. Nous, 44(2), 224–244
Reed, B. (2013). Practical matters don’t affect whether we know. Contemporary debates in epistemology. (2nd ed.). Wiley-Blackwell.
Roeber, B. (2020). How to argue for pragmatic encroachment. Synthese, 197, 2649–2664
Rysiew, P. (2011). Epistemic Contextualism. The Stanford Encyclopedia of Philosophy (Winter 2011 Edition), E. N. Zalta (ed.) http://plato.stanford.edu/archives/win2011/entries/contextualism-epistemology/.
Schaffer, J. (2006). The irrelevance of the subject: Against subject-sensitive invariantism. Philosophical Studies, 127, 87–107
Schroeder, M. (2012). Stakes, withholding and pragmatic encroachment on knowledge. Philosophical Studies, 160, 266
Sosa, E. (2015). Judgment and agency. Oxford University Press.
Stalnaker, R. (1999). Context and content. Oxford University Press.
Stanley, J. (2005). Knowledge and practical interests. Oxford University Press.
Stanley, J. (2008). Knowledge and certainty. Philosophical Issues, 18(1), 35–57
Turri, J. (2017). Epistemic contextualism—An idle hypothesis. Australasian Journal of Philosophy, 95(1), 141–156
Weatherson, B. (2005). Can we do without pragmatic encroachment? Philosophical Perspectives, 19, 417–443
Williamson, T. (1996). Knowing and asserting. Philosophical Review, 105, 489–523
Williamson, T. (2000). Knowledge and its limits. Oxford University Press.
Williamson, T. (2005). Contextualism subject-sensitive invariantism, and knowledge of knowledge. Philosophical Quarterly, 55, 213–235
Acknowledgements
This research project (with Nikola Kompa) was financially supported by the State of Lower Saxony, Hannover, Germany. I wish to thank my research assistants Daniel Giron, Zeev Goldshmidt, Nir Gottlieb and Ofer Zak. I wish to thank for valuable comments and encouragement Christian Beier, Sandy Goldberg, Andy Egan, Tyler Burge, Hartry Field, Matt McGrath, Hans Rott, and especially Tim Williamson
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This article is part of the topical collection on Epistemic Significance of Non-Epistemic Factors, edited by Nikola Kompa, Igal Kvart, and Andrea Robitzsch.
Rights and permissions
About this article
Cite this article
Kvart, I. Resolving Bank-Type Puzzles via Action-Directed Pragmatics. Synthese 200, 298 (2022). https://doi.org/10.1007/s11229-022-03648-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s11229-022-03648-4
Keywords
- Pragmatic encroachment
- Epistemic contextualism
- Action-directed pragmatics
- SSI
- Bank case
- Knowledge
- Truth-conditions
- Assertibility
- Sayability
- Assertibility conditions
- Stakes
- Ascriber
- Context
- Epistemic position
- Overall sayability
- Epistemic sayability
- Norms
- Instrumental norms
- Epistemic norms
- DeRose
- Stanley
- McGrath
- Fantl
- Weatherson
- Pragmatic functors
- Negative polarity
- Invariantism
- Steering Thrust