Misunderstanding and partial understanding are pervasive and inevitable phenomena, and attributions of content despite them are an integral part of common practice. Burge (Reference Burge1979: 102)
1. Introduction
Skeptical invariantism (hereafter SI) is, roughly, the family of theories of knowledge that claim that the meaning of ‘know’ is invariant and that the strength of justification required for knowledge is so demanding that we know little (if anything at all).Footnote 1 Skeptical invariantists often understand the strength of justification required for knowledge in terms of deduction and propose that, in order to really know, justification must entail the truth of belief, or make the belief absolutely certain.Footnote 2
Unsurprisingly, SI is a theory that is widely considered implausible and lies at the margins of the current debate about the semantics of knowledge – and perhaps rightly so. But various epistemologists – with the exception of Hawthorne (Reference Hawthorne2004) – have relied on a single, often quickly aired, argument for their rejection of the viewFootnote 3: what we may call ‘the argument from semantic awareness’.Footnote 4 Roughly, ‘the argument from semantic awareness’ indicates that ordinary agents are not aware of the meaning of ‘know’ that SI proposes. Given that the semantic intuitions of ordinary agents are generally reliable, this implies that SI is implausible as a theory of ‘know’. Therefore, SI is implausible and should be rejected.
In this paper, I focus on the stronger extant formulation of ‘the argument from semantic awareness’ (offered by Dinges Reference Dinges2015) and explore how SI could, in principle, be rendered coherent with the argument (even if SI is not to be considered overall plausible). To this effect, I suggest an overlooked semantically externalist model of meaning and semantic awareness of ‘know’ that renders SI coherent with ‘the argument from semantic awareness’. The model is in the spirit of the paradigmatic semantic externalism of Putnam (Reference Putnam1975/1997), Burge (Reference Burge1979) and Kripke (Reference Kripke1981).
The goal of the paper is modest. It is not to defend, let alone vindicate SI, but to indicate that SI is coherent with ‘the argument from semantic awareness’ in light of an overlooked externalist account of meaning and semantic awareness. If this is to the right direction, the interesting conclusion we arrive at is that we cannot reject SI on the sole basis of ‘the argument from semantic awareness’. Rather, we should reject SI only if it is overall less explanatorily powerful than theoretical alternatives and this is a contentious assertion we cannot even begin to inquire into here.Footnote 5
It may be objected that the modest conclusion of the paper is almost trivial to bother to argue for, but the response would be that it is worthwhile because it is often thought that SI is singlehandedly defeated by ‘the argument from semantic awareness’. As the paper shows, it is not. It might also be objected that SPI is not a very good theory in terms of explanatory power and, therefore, it is quite unlike good scientific theories that are explanatorily powerful and their key concepts are subject to externalist treatment (dolphin, heat, salt, water, tomato etc.). But, for the sake of argument, we will remain neutral on the question of the overall plausibility of SPI and, therefore, the objection would be dialectically uncharitable to the argument of the paper that modestly seeks to establish the mere coherence of SPI with ‘the argument from semantic awareness’.
Besides, at least some competent epistemologists have argued at some length for the explanatory power of SI in regard to some central epistemic puzzles (Gettier, lottery and value problems, dogmatism and preface paradoxes, concessive knowledge attributions, DeRose's bank and Cohen's airport cases etc.) and it would be dogmatic to simply ignore their argued views, especially in light of widespread philosophical disagreement about ‘know’ and the plethora of extant theories (invariantist, contextualist, relativist, expressivist etc.).Footnote 6 Given that overall assessment of SPI is something that we cannot venture into here, the principle of dialectical charity requires that we at least leave open the possibility that SPI could be explanatorily powerful.
With these two preliminary objections set aside, let us first introduce Dinges’ (Reference Dinges2015) version of the argument.
2. Dinges’ (Reference Dinges2015) argument from semantic awareness
As mentioned, the stronger formulation of ‘the argument from semantic awareness’ has recently been offered by Dinges (Reference Dinges2015). Innovatively, Dinges (Reference Dinges2015: 2590–1) introduces a subtle linguistic device that tests in linguistic practice our semantic intuitions about ‘know’, but his linguistic device is not subject to objections that linguistic devices of previous statements of the argument meet.Footnote 7 He finds that our semantic intuitions about ‘know’ defeat SI because they indicate that we are not aware of the meaning of ‘know’ that SI proposes.
Before we introduce Dinges’ linguistic test, we need to unpack Dinges’ (Reference Dinges2015: 2579) elegant explication of the semantic workings of a version of SI that he finds to some extent explanatorily fruitful (e.g. it accounts for DeRose's bank cases), namely, skeptical pragmatic invariantism (hereafter, SPI). He does so by reference to the notion of substitutional implicatures. That is, ‘implicatures where the speaker doesn't convey what is said but something else instead’. According to Dinges’ explication of SPI, knowledge attributions of the form ‘S knows that p’ literally say one (infallibilist) thing but due to loose use (via substitutional implicature) mean another (fallibilist) thing that is close enough to what is said. The literal content is the semantic content of ‘S knows that p’ while the implicated content is the pragmatic content.
Dinges (Reference Dinges2015: 2580) understands what it is to be close enough (via substitutional implicature) to what is said in the following way: ‘in order to be close enough to knowing that p, one must satisfy all the conditions for knowledge … except the justification condition … S is close enough (for the purposes of the low standards case) to knowing that p iff p, S believes that p and S can rule out all likely alternatives to p’ (his own emphasis). According to SPI, the justification condition is not often satisfied because in order to really know p justification must be deductive-infallibilist justification and not merely fallibilist (and this is not something that happens very often). In what follows, Dinges’ (Reference Dinges2015) explication of SPI is assumed as our preferred ‘toy’ SI theory.
In order to formulate his argument, Dinges (Reference Dinges2015: 2588) stipulates – what he calls – ‘the awareness principle’, that is, a principle that governs the expression of substitutional implicatures in ordinary discourse.Footnote 8 As he says, ‘in the case of substitutional implicatures (i.e. the implicatures relevant for SPI, that substitute what is literally said), people are aware of the discrepancy between what they say and what they implicate’ (Dinges Reference Dinges2015: 2588). To substantiate the principle he offers some linguistic evidence by means of three intuitive examples of ordinary substitutional implicatures. These are the following:
(1)
A: The plane was a mile long.
B: That's absurd! No plane is a mile long!
He notes that ‘[i]n a suitable context, A's first utterance generates a substitutional implicature to the effect that the plane was huge. B's response, however, is an objection only to the falsity of the literal content of A's claim. Correspondingly, A, qua competent speaker, will straightforwardly realize that B is missing the point (intentionally or unintentionally). Thus, A can plausibly respond by saying, ‘A: Oh c'mon, you know what I meant!’. Similar things hold for the other two examples:
(2)
A: It's three o'clock.
B: It's one minute past!
A: Oh, c'mon. You know what I meant.
(3)
A: France is hexagonal.
B: But what about this bump here and this bump there?
A: Oh, c'mon. You know what I meant!
Dinges (Reference Dinges2015: 2589) concludes that ‘[t]he awareness principle … seems to make the right predictions’. With these intuitive examples in mind, he proceeds (Reference Dinges2015: 2590–1) to propose what he calls ‘the awareness objection’ to SPI, which he explicates in terms of the following intuitive example:
(D1)
DeRose: I know that the bank will be open.
His wife: I doubt that. The bank could have changed its hours.
#DeRose: Oh, c'mon. You know what I meant!
As Dinges (Reference Dinges2015: 2590) observes, ‘[t]he above, clarifying response … seems unnatural. This is the awareness objection to SPI.’ (Dinges’ own emphasis). That is, SPI is problematic because if the assumed theory of ‘know’ in (D1) is SPI, and the DeRose character is supposed to be aware of the substitutional implicature expressed to the effect that not all logical error-possibilities have been neutralized, the clarifying retort ‘#DeRose: Oh, c'mon. You know what I meant!’ would have been felicitous to be said because it would clarify that the knowledge assertion was loose and only expressing approximation.
The problem for SPI is that this is not what typically happens in ordinary discourse because it is a rather infelicitous thing to say. According to Dinges (Reference Dinges2015: 2590–1), this suggests that the DeRose character (and ordinary speakers) are not aware of the substitutional implicature (i.e. what is meant) and its discrepancy with what is said. Hence, SPI is mistaken on empirical linguistic grounds because it flouts the awareness principle and, ‘in the absence of a plausible response, the view should be rejected after all’ (Dinges Reference Dinges2015: 2592).
This brief presentation sums up Dinges’ (Reference Dinges2015) version of ‘the argument from semantic awareness’. In the next section we raise two preliminary externalist qualms about the argument. They are meant to indicate that SPI is, in principle, coherent with ‘the argument from semantic awareness’ if semantic externalism is assumed and, thereby, suggest the possibility of an externalist response to ‘the argument from semantic awareness’. We substantiate this possibility in the rest of the paper.
3. Externalist qualms about ‘the argument from semantic awareness’
The first qualm is a methodological one. Ordinary semantic intuitions are not to be considered rebutting defeaters to explanatorily promising theories (philosophical and scientific). That is, unless we accept a dubious methodological principle that prescribes saving ordinary intuitions/appearances at all cost, ordinary semantic intuitions are not to be considered rebutting defeaters to explanatorily promising theories.Footnote 9 Besides, it is well-known that often our semantic intuitions misguide us about true semantic content. We are all familiar with semantic externalism and Putnam's (Reference Putnam1975) famous ‘meaning ain't in the head’ remark. So, ‘the argument from semantic awareness’ should not be considered a rebutting defeater (as has been considered by most proponents) but an undercutting defeater.
The upshot of the first qualm is that the linguistic phenomena Dinges (Reference Dinges2015) points to are, in principle, coherent with SPI. Were SPI to be independently explanatorily powerful, as some epistemologists contend, the awareness objection would not amount to a rebutting objection (unless we assume a dubious methodological principle). Ordinary speakers could be flouting the awareness principle all the time because they are partially semantically blinded about ‘know’, while SPI is nonetheless true.
A second qualm (that corroborates our above diagnosis) is that, were Dinges (Reference Dinges2015) to be right about the awareness objection, rejection of theories would over-generalize to good scientific theories. This would be the case because the linguistic fact he is drawing attention to it is not without precedent, or especially surprising in the history of science. It happens routinely with empirical scientific concepts, but we do not reject explanatorily powerful scientific theories just because of the prescriptions of the awareness principle. We simply accept that ordinary speakers are semantically blinded (at least partially) and cling to the theory. By parity of reasoning, the same should hold for SPI, were it ever to be shown to be of independent explanatory power (as some epistemologists contend, e.g. Fogelin (Reference Fogelin1994), Bonjour (Reference Bonjour2010), Dodd (Reference Dodd2011) and even Dinges (Reference Dinges2015) himself).
Ordinary speakers may not be aware that water is H20, that arthritis is an ailment of the joints (and not muscles), that time is physical-frame-relative (and not Newtonian-absolute), that tomatoes and eggplants are fruit (and not vegetables), but this is not to refute the respective scientific theories. To think otherwise, it is to subscribe to strong semantic internalism about meaning and, as Putnam (Reference Putnam1975/1997), Burge (Reference Burge1979) and Kripke (Reference Kripke1981) warned us, meaning is not simply ‘in the head’. It is also contributed by external reality, be it physical, chemical, biological, or other. Perhaps then the same holds with the case of knowledge.Footnote 10 Ordinary speakers may not be aware that knowledge is infallible, deductively justified true belief but this is not to refute the respective philosophical theory, namely, SPI.
Here is an example to illustrate the semantic point drawn above:
A: Flipper the dolphin is a fish.
B: Dolphin but a fish? I doubt that.
# A: Oh, c'mon, you know what I meant.
According to a semantically externalist interpretation of this short exchange, what it is literally said in A is that ‘Flipper the dolphin is a fish’, which is strictly speaking false because dolphins are mammals, not fish. Loosely speaking, via a substitutional implicature, what is meant in A is something close enough to what is said, namely, that Flipper the dolphin exemplifies prototypical properties of fish i.e. creature of the sea, swims, has a fin etc., but still it is not really a fish. When B objects that what is literally said by A is dubious, #A's clarifying response suggests that what was meant all along was the content of a substitutional implicature.
#A's clarifying response, however, is infelicitous unless A is aware of the technical categorization of dolphin (i.e. marine mammal, not fish) on the basis of biological theory. But this empirical linguistic fact would hardly convince biologists to abandon mammal theory. The same should hold with epistemologists, if SPI is ever shown to be independently plausible, which is a question not to be opened here.
It may again be objected that the externalist semantic interpretation of the exchange is mistaken because obviously A does not express any substitutional implicature with ‘Flipper the dolphin is a fish’ and, therefore, this example is disanalogous to Dinges’ own examples (time, hexagonality) where a substitutional implicature is obviously expressed. As it is disanalogous, the dolphin example fails to undermine Dinges’ ‘argument from semantic awareness’.
The objection, however, misses the externalist point drawn because it begs the question of semantic interpretation (internalist or externalist) against externalism (and the case for SPI). The objection insists on a semantically internalist interpretation of the exchange that ignores the possibility of a substitutional implicature being expressed. This is the case because the objection sticks to a semantically internalist interpretation that refuses to accept that semantic externalism might be correct and that we might be at least partially semantically blinded to the true meaning of the used concept.Footnote 11
In other words, along externalist lines, we could be using sentences without being aware that all we are expressing is a substitutional implicature and not the externally fixed literal content. Besides, ordinary agents are not very sensitive to the technicalities of semantic theory (e.g. the semantics/pragmatics distinction) and it should not be surprising if they fail to distinguish between expressing a substitutional implicature and not the externally fixed literal content.Footnote 12
Thus, the externalist interpretation of the exchange is at least as coherent as the internalist is and cannot be brushed aside on pain of begging the question against externalism (and by extension the case for SPI). So the dolphin example still stands and by analogy prima facie provides an externalist semantic model for ‘know’ that could, in principle, rescue it from ‘the argument from semantic awareness’.
The rejoinder might be that Dinges’ intuitive examples (time, hexagonality) still pose a challenge because they outline an intuitive, internalist model of meaning and semantic awareness that is quite unlike the externalist model implied by the dolphin example. If the externalist model is to be taken seriously, it should offer at least an equally plausible interpretation of these examples with the internalist interpretation. This is a reasonable concern and we return to it in section 5 to argue that Dinges’ examples (time, hexagonality) can also be subjected to an externalist interpretation (and a more nuanced one at that).
The above externalist qualms indicate that Dinges’ (Reference Dinges2015) awareness objection uncritically relies on semantic internalism and that SPI is, in principle, coherent with the awareness objection if we assume semantic externalism. In the last section we substantiate this conceptual possibility and offer a linguistically sensitive, externalist account of meaning and semantic awareness about ‘know’. For time being, in the next section we further bolster the externalist line of argument and make explicit the semantically internalist underpinnings of Dinges’ (Reference Dinges2015) awareness principle (upon which the awareness objection hinges). For the sake of clarity, we also explicate our understanding of the awareness principle in terms of an understanding of the semantics/pragmatics distinction.
4. The semantic internalism of the awareness principle
As we saw, Dinges (Reference Dinges2015) implicitly understands the awareness principle in broadly internalist terms. But his formulation of the principle is actually ambiguous between semantic internalist and semantic externalist readings. The ambiguity is evident in the way Dinges (Reference Dinges2015: 12) stipulates the awareness principle itself. He writes: ‘in the case of substitutional implicatures (i.e. the implicatures relevant for SPI, that substitute what is literally said), people are aware of the discrepancy between what they say and what they implicate’. Note that initially he talks about the impersonal what is said/what is meant, but immediately after he shifts and talks in terms of personal, third-person plural about what they say and what they implicate.
The two ways of talking are not necessarily semantically equivalent because they bear different semantic implications. The impersonal what is said/what is meant points towards an externalist reading of the awareness principle because it suggests that meaning is independent of social-conventional usage.Footnote 13 The personal, third-person plural what they say/what they implicate points towards an internalist reading of the awareness principle because it suggests that meaning is dependent on social-conventional usage. Let us distinguish between these two readings with regard to the awareness principle. These are the following:
(Inter) Ordinary speakers are aware of the discrepancy between what is conventionally said (by speakers) and what is conventionally meant (by speakers).
(Exter) Ordinary speakers are aware of the discrepancy between what is really said and what is conventionally said (by speakers) and conventionally meant (by speakers).Footnote 14
Let us briefly explain. What is conventionally said and conventionally meant is what is typically said and meant by ordinary speakers in accordance with their extant sociolinguistic convention. It is what expressions have come to say and mean in their sociolinguistic convention due to established Wittgensteinian, pragmatic regularities of use (cf. Bach Reference Bach2008: 71).Footnote 15 As what is conventionally said and conventionally meant tends to be introspectively accessible to speakers (due, at least partly, to internalization of lexical meaning through socialization from early childhood),Footnote 16 this is an internalist awareness principle. The internalist reading is the reading assumed by Dinges’ awareness principle.
To illustrate, take Dinges’ (Reference Dinges2015) time example:
A: It's three o'clock.
B: It's one minute past!
A: Oh, c'mon. You know what I meant.
According to Dinges’ (Reference Dinges2015: 2589) implicitly internalist interpretation of the exchange, what is said and what is meant is conventionally fixed by established pragmatic regularities of use at the social level.Footnote 17 That is, in everyday life, we tend to use temporal expressions loosely and in approximation because it is practical to do so. According to Dinges, this is what is (conventionally) meant (via a substitutional implicature). In everyday life, we also tend to assume that there is an absolutely accurate time that temporal expressions purport to represent and that this is the literal content. According to Dinges, this is what is (conventionally) said. Thus, Dinges implicitly assumes the internalist reading of the awareness principle.
In contrast, according to an externalist interpretation of the exchange, what is really said is what is to be said in accordance with our best theory (philosophical or scientific) of the meaning of an expression. It is what expressions really say according to our best available theory, which is often not conventionally fixed by established pragmatic regularities of use. As what is really said (according to our current best theory) need not be introspectively accessible to ordinary speakers (who are mostly not aware of various corresponding, often technical theories and their implications), this is an externalist awareness principle. In the next section, I apply the semantically externalist interpretation of the awareness principle to Dinges’ time and hexagonality examples.
A simple example to illustrate how the two different readings of the awareness principle are supposed to work is ‘water’. With ‘water’ we conventionally say and conventionally mean something like ‘the potable liquid that is odorless, tasteless, transparent, fills our rivers, quenches our thirst’ etc.Footnote 18 What is conventionally said and conventionally meant coincide in the case of ‘water’, although in other cases, such as the time example, do not.
But what is really said by ‘water’ is that it is ‘the natural kind with molecular structure H2O’, and ordinary speakers often are not aware of this because they are not aware of the chemical fact (and molecular theory, elements theory).Footnote 19 When they are not aware of the chemical fact, ordinary speakers with ‘water’ conventionally say and conventionally mean ‘the potable liquid that is odorless, tasteless etc.’ (via substitutional implicature), although what is really said with ‘water’ is ‘the natural kind with molecular structure H2O’.Footnote 20
The identified ambiguity of the awareness principle is important because, as we have seen with the dolphin and water examples, ordinary speakers are generally comfortable with the internalist reading but not so comfortable with the externalist reading of the awareness principle (although SPI need not be rebutted by this psycholinguistic fact). If ordinary speakers are only aware of what is conventionally said and conventionally meant and not aware of what is really said (and the discrepancy with what is conventionally said and conventionally meant), this does not defeat an explanatorily powerful theory. We do not (and should not) reject good theories (physical, biological, chemical, philosophical etc.) just because of the semantic intuitions of ordinary speakers, even if we have to attribute some semantic blindness to speakers, which is inevitably counter-intuitive.Footnote 21
To help us clarify the emerging externalist picture of the semantic workings of ‘know’ (and other lexical items that may subscribe to the same model), we should explicate the drawn semantic internalism/externalism awareness principle in terms of the fundamental semantics/pragmatics distinction.Footnote 22 Situating our externalist understanding of the semantics workings of ‘know’ (and other lexical times) within the perspective of the semantics/pragmatics distinction will help us better understand the complexities of knowledge discourse (and other lexical items discourse). Of course, the semantics/pragmatics distinction is itself a very controversial distinction, but for the sake of clarity we cannot afford to completely ignore it.Footnote 23
For current purposes, we can rely on a fairly orthodox understanding of the distinction for both linguists and philosophers, although we will move beyond it in order to incorporate the externalist semantic lesson. According to a standard textbook explication, semantics is the province of literal linguistic meaning, of what is being said. Pragmatics is the province of non-literal non-linguistic meaning, of what is being meant. As is well-known, the two domains of meaning might pull apart because we might not mean things we explicitly say (i.e. in the mouth of a drug dealer, saying ‘Bring the fish’ might mean bring the illegal cargo) and might not explicitly say things we clearly mean (i.e. the Gricean ‘John is a competent philosopher, he has a special talent in calligraphy’, implying philosophical incompetence).
This is a standard way of drawing the semantics/pragmatics distinction in mainstream truth-conditional semantic frameworks (cf. Kearns Reference Kearns2000; Fromkin et al. Reference Fromkin, Rodman and Hyams2011), but it comes with an externalist twist. Roughly, semantics is understood to be fixed by the truth-conditions the linguistic meaning confers, while pragmatics is understood to be fixed by the contextually informed actual linguistic usage. If the linguistic meaning is fixed by external factors then the truth-conditions are externally fixed as well. In this sense, semantics deals with the true (or real) meaning suggested by our current best theory that is supposed to describe, as best as possible, reality. This is semantics proper.
Pragmatics is understood to be fixed by the contextually informed actual linguistic usage. This, however, indicates that we should distinguish between what is conventionally said and what is conventionally meant. In examples like the time case or the hexagon case, there might be a distinction between what ordinary speakers think the concept really says (as internalized by convention) and how they actually apply the concept for all practical purposes. We may appropriate a distinction from Lycan (Reference Lycan2006: 164–5) and call the first pragmatics domain of meaning semantic pragmatics and the second pragmatics domain pragmatic pragmatics.
As I understand it, semantics proper is normative: it deals with how we, in principle, ought to use concepts, propositions, phrases (according to standards of semantic correctness, such as extension or accurate reference). Pragmatics is descriptive: it deals with how we actually use the concept of knowledge, where often actual use is informed by conventional pragmatic regularities of use (that may have less to do with standards of semantic correctness and more to do with the practical exigencies of everyday life). If this threefold distinction is to the right direction, to understand the semantic workings of ‘know’ (and other lexical items) we need to understand how discourse operates at the interface of the semantics and the pragmatics (along with the external semantic contribution of our best science and philosophy).
With this amount of semantic clarification under our belt, let us revisit Dinges’ (Reference Dinges2015) examples and apply semantic theory to practice. First come his non-epistemic examples.
5. Back to Dinges’ (Reference Dinges2015) non-epistemic examples
Our discussion so far has strived to expose the semantically internalist underpinnings of the awareness principle and objection articulated by Dinges (Reference Dinges2015) and raise externalist qualms about them that indicate that SPI could, in principle, be coherent with the objection. This opens conceptual space for the possibility of SPI, but unless we substantiate the possibility by means of a linguistically sensitive, externalist treatment of ‘know’, the mere possibility of SPI does not seriously weaken ‘the argument from semantic awareness’. We pursue this task in the next and final section.
Before we move on to this task, we are challenged to consider Dinges’ (Reference Dinges2015) non-epistemic examples of semantic awareness and examine whether these can be treated in an equally plausible externalist way. If not, this will not by itself defeat the possibility for an externalist treatment of ‘know’ (and SPI), but it will indicate that an externalist semantic treatment might not be very widely applicable. In response to this challenge, in this section we attempt to offer an externalist treatment of Dinges’ (Reference Dinges2015) non-epistemic examples.
As we have seen, in his intuitive examples (time, hexagonality) Dinges (Reference Dinges2015) had no problem in explaining how speakers naturally understand what was (conventionally) said and what was (conventionally) meant (according to an internalist interpretation). The challenge then is to explicate in an at least equally plausible, externalist way the semantic workings of the intuitive examples Dinges has provided.
Indeed, an externalist interpretation of Dinges’ examples is available and actually offers a more nuanced semantic picture than the internalist interpretation. That is, it might be that even in Dinges’ own examples the speakers are not aware of what is really said but are only aware of what is conventionally said and conventionally meant. Hence, Dinges’ (Reference Dinges2015) examples might in fact be better explained by the ambiguous formulation of the awareness principle and, therefore, support the externalist case. To support our criticism of the awareness principle, let us illustrate via Dinges’ (Reference Dinges2015) own examples:
A: France is hexagonal.
B: But what about this bump here and this bump there?
A: Oh, c'mon. You know what I meant!
According to an externalist interpretation of the exchange, when A says ‘France is hexagonal’ what he conventionally means (via a substitutional implicature) is that France is six-sided and six-angled in a way that approximates (or is close enough) to a hexagon. What A conventionally says is that France is an exact hexagon, which is strictly speaking false, of course, because it is not. He is thus competent with the conventional meaning (what is conventionally said/what is conventionally meant) of hexagonality. But A does not understand what hexagonality really says because he is not an expert geometrician.Footnote 24 No doubt, it would be absurd to expect such performance in ordinary discourse.Footnote 25 It would also be absurd to think that the agent is aware of geometrical theory (Euclidean, non-Euclidean) and what it implies for hexagonality. Thus, the speaker A is not transparently aware of what is really said. As we shall see, analogous things could obtain in Dinges’ (Reference Dinges2015) epistemic example as well and ultimately undermine it. The same holds with the time example:
A: It's three o'clock.
B: It's one minute past!
A: Oh, c'mon. You know what I meant.
According to an externalist interpretation, A is minimally competent with the meaning of time (e.g. the one-directionality of time, that it applies to events etc.). When A says ‘It's three o'clock’ he conventionally means (via substitutional implicature) that it is approximately (or close enough to) three o'clock. What A conventionally says with ‘It's three o'clock’ is that it is exactly three o'clock.
But A does not understand what time really means (according to our current best theories, anyway) because he is not an expert (meta)physicist of time.Footnote 26 It would be absurd to expect such performance from speakers in ordinary discourse. It would also be absurd to think that the agent is aware of relativity theory and related conceptual issues in metaphysics (e.g. McTaggart's famous paradox) and what they imply about time. Thus, the speaker A is not aware of what is really said.Footnote 27
The externalist interpretation of Dinges’ (Reference Dinges2015) non-epistemic examples suggests a revised weak awareness principle that would save the best from both internalist and externalist readings of Dinges’ awareness principle and would be making the right predictions in regard to the canvassed examples. On the one hand, the weak awareness principle would be explaining (with the internalist reading) how ordinary agents tend to be aware of the conventional content of what they say and mean. On the other hand, the weak awareness principle would be explaining (with the externalist reading) how ordinary agents often tend to miss what is really said. This awareness principle is provisionally the following:
(WeakAw) Ordinary speakers tend to be aware of what is conventionally said and conventionally meant (due to internalization of established pragmatic regularities of use within their sociolinguistic convention). But ordinary speakers often tend to miss what is really said (according to our current best theories, anyway) and by extension the discrepancy between conventional-pragmatic content and real-semantic content.
A further attraction of the weak awareness principle is that it allows that, even if we are not actually aware of what is really said, we can potentially become aware of it. There is no reason why someone could not, in principle, learn about the (meta)physics of time and its implications about time, geometrical theory and its implications about hexagonality and so on. The same could hold about the theory of knowledge and its implications about ‘know’. If SPI is shown to be independently plausible, which is a question not to be opened here, we could learn about SPI and its implications about ‘know’.
6. Back to Dinges’ (Reference Dinges2015) epistemic example
We have so far exposed the implicit internalist underpinnings of Dinges’ awareness principle and objection and suggested an externalist interpretation of Dinges’ non-epistemic examples that takes into account externalist semantic insights along the lines of the weak awareness principle. We are now poised to revisit Dinges’ (Reference Dinges2015: 14) epistemic example (D1) and place it within the developed externalist perspective of SPI and the weak awareness principle. Recall the example:
(D1)
DeRose: I know that the bank will be open.
His wife: I doubt that. The bank could have changed its hours.
#DeRose: Oh, c'mon. You know what I meant!
As we concurred with Dinges, this is typically infelicitous for ordinary speakers because the clarifying response is rather unnatural (at least for low standard cases). Although SPI suggests that we express (via substitutional implicature) a weak fallibilist concept of knowledge that does not eliminate far-fetched, seemingly irrelevant possibilities of error, the clarifying response to the doubting of the literal, infallibilist content (that misses the point) seems rather infelicitous, which indicates that the DeRose character (and ordinary speakers) is not aware of the discrepancy between what is said and what is meant, according to SPI. That was the initial problem.
But as we have argued, ordinary speakers are not typically expected to be aware of what is really said. They are not expected to have such high-grade semantic awareness. Ordinary speakers are expected to be aware of what is conventionally said and conventionally meant. If this is to the right direction, the widespread conventional expression of the fallibilist concept in knowledge assertions/attributions (via substitutional implicatures) need not imply that ordinary speakers have to be aware of the content of what is really said. They might be minimally competent with ‘know’ (e.g. that it entails truth, belief, justification),Footnote 28 but a deep background theory is required for the task of recognizing that we fall prey to partial understanding about knowledge.Footnote 29
The cognitive fact that deep background theory is required for the task of recognizing that we fall prey to partial understanding about various concepts in ordinary discourse should not come as a surprise. Given cognitive penetration of the semantic content of words, expressions and sentences by background theory this should be expected. The meaning of words and sentences is often interactive with and penetrable by the deep background theory we have.Footnote 30 ‘Mass’ for example meant something different in Newton's mouth than in Einstein's mouth. ‘Caloric’ and ‘phlogiston’ were considered extensional and meaningful before ‘heat theory’ and ‘oxygen theory’ but not afterwards.Footnote 31 The same goes with ‘fish’ in the mouth of a biologist, ‘water’ in the mouth of a chemist and ‘fish’ and ‘water’ in the mouth of a layperson.
Here is another example:
A: Tomatoes are my favorite vegetable.
B: But wait, tomatoes are fruit, not vegetable.
#A: Oh, c'mon. You know what I meant.
According to an externalist interpretation, unless A is aware of fruit theory and that, strictly speaking, tomatoes are fruit and not vegetable, the clarifying response would seem again infelicitous.Footnote 32 Of course, typically for practical reasons we need not be such nit-picky ‘sticklers for correct speech’ (cf. Grice Reference Grice1989: 45) and make our ordinary thought and talk so cumbersome. To the extent that it does not fail our practical goals, we can speak loosely. Analogous things could happen with, and eventually undermine, Dinges’ own epistemic example against SPI. Recall:
(D1)
DeRose: I know that the bank will be open.
His wife: I doubt that. The bank could have changed its hours.
#DeRose: Oh, c'mon. You know what I meant!
According to an externalist, SPI interpretation of the exchange, when A says ‘I know that the bank will be open’, he conventionally says and conventionally means (as in the case of ‘water’, they coincide here) the substitutional implicature that he fallibly knows that the bank will be open in a way that approximates (or is close enough) to (infallible) knowledge. He is thus competent with the conventional, fallibilist meaning of ‘know’. But he does not understand what ‘know’ really says because he is not aware that SPI is true (assuming that it is true, for the sake of argument).
As with the dolphin, water and tomatoes examples, assuming that SPI is true, if the DeRose character is not aware of SPI theory and the implication for ‘know’ in low standard cases, then ‘A: Oh, c'mon. You know what I meant!’ makes little sense. DeRose's knowledge assertion really says one (infallibilist) thing, but due to loose use (via substitutional implicature) it conventionally says and conventionally means another (fallibilist) that is close enough to what is really said. But the DeRose character (and ordinary speakers) lack the high-grade semantic awareness to understand the discrepancy between what is really said and what is conventionally said and conventionally meant that would make felicitous the clarifying response. He is thus partially semantically blinded.
But although the DeRose character (and ordinary speakers) are partially semantically blinded about ‘know’ if SPI is true, they are not in a linguistically hopeless situation. They do have subtle linguistic devices at their disposal that can help them disambiguate what concept of ‘know’ (fallibilist or infallibilist) is expressed in such (and other) contexts.Footnote 33
To disambiguate the concept of knowledge expressed (and rehabilitate sense in Dinges’ (Reference Dinges2015) example) we could either invoke some linguistic device that clarifies to otherwise competent ordinary speakers that the relaxed fallibilist concept is conventionally said and conventionally meant (via substitutional implicature), or be an expert epistemologist that understands straightaway (due to cognitive penetration) that the fallibilist concept of knowledge is conventionally said and conventionally meant. Both options are available for SPI. Let us examine the former option first.
We could appeal to doxastic attitude verbs such as ‘think’ and ‘believe’ to clarify to ordinary speakers that we are expressing the fallibilist concept.
*(D1)
DeRose: I think I know that the bank will be open.
His wife: I doubt whether you know it. The bank could have changed its hours.
#DeRose: Oh, c'mon. You know what I meant!
The doxastic attitude verb ‘think’ (with the proper emphasis) takes as a complement the knowledge assertion that implicates that the speaker acknowledges some relatively salient error-possibilities that are, however, remote enough in logical space to be irrelevant for current purposes. Therefore, they do not threaten the knowledge assertion (e.g. change in hours) and they can be ‘properly ignored’ (in Lewis’ Reference Lewis1996 words).Footnote 34 DeRose thus makes explicit that he thinks that it is likely that his knowledge assertion would be true and a case of knowledge (by low fallibilist standards).
The DeRose's wife doubting challenge comes then all too naturally to DeRose because he has made explicit that there is some relatively salient but remote and thereby irrelevant possibility that he is mistaken (by low standards). Thus, the #DeRose clarifying response ‘Oh, c'mon. You know what I meant!’ comes as a rather natural response to the doubting challenge. It clarifies that he means that in spite of some relatively salient but remote and thereby irrelevant for current purposes error possibilities, he expects the assertion to be true and a case of knowledge by low standards (and not that he is absolutely certain that his assertion will be true and an instance of knowledge, even by high infallibilist standards).
Other linguistic devices like modifiers (i.e. adverbs, adjectives) could help with making explicit the expression of the infallible concept.Footnote 35 To illustrate consider the following revised *(D1) exchange:
**(D1)
DeRose: I really know that the bank will be open.Footnote 36
His wife: I doubt that. The bank could have changed its hours.
#DeRose: Oh, c'mon. You know what I meant!
In this example the modifying adverb ‘really’ makes explicit that the infallibilist concept is expressed and the ‘#DeRose: Oh, c'mon. You know what I meant!’ response comes out as infelicitous because it is semantically redundant. The adverb ‘really’ has already clarified what concept of knowledge is used and, therefore, there is no need for any further clarification. If we apply Grice's razor – ‘Do not multiply senses between necessity’ – and remove the redundant ‘#DeRose: Oh, c'mon. You know what I meant!’ response, the exchange becomes perfectly natural. Compare:
**(D1)
DeRose: I really know that the bank will be open.
His wife: I doubt that. The bank could have changed its hours.
So, there are linguistic devices that allow us to distinguish which concept of knowledge is expressed (fallibilist or infalibilist) in a certain context – even if ordinary speakers never heard of the technical fallibilism/infallibilism distinction. Let us now examine the latter option of resolving the problem. Compare:
***(D1)
DeRose: I *[fallibly]* know that the bank will be open.
His wife: I doubt that. The bank could have changed its hours.
#DeRose: Oh, c'mon. You know what I meant!
With ‘*[fallibly]* know’ I mean that the speaker consciously employs the fallibilist concept of knowledge (without articulating *[fallibly]*) because he is an expert epistemologist (like most readers of this piece, I presume). As an expert, his understanding of background theory of knowledge cognitively penetrates the content of ‘know’ and can grasp the fallibilist concept of knowledge supposed to be expressed in this low standards context. Plausibly, such cognitive ability is not often to be found in non-experts, common folk that are lacking in deep background epistemological theory (and the fallibilism/infallibilism distinction).Footnote 37
As we saw, the lack of this cognitive ability was the source of the initial awareness problem. But once the expert is aware of the concept in operation and that it does not eliminate all possibilities of error but only relevant ones for current purposes, then it all makes sense. The clarifying response (‘Oh, c'mon. You know what I meant!’) to the doubting challenge comes out as felicitous because it clarifies that the speaker uses the weak fallibilist concept of knowledge and not the infallibilist concept, which is the correct one according to SPI. At least if we assume that DeRose's wife also counts as an expert in epistemological theory (and competent with the fallibilism/infallibilism distinction) and can follow DeRose's clarifying response.
Overall, I conclude that the linguistic evidence that Dinges (Reference Dinges2015) has drawn attention to is, in principle, coherent with SPI in light of semantic externalism. What is more, Dinges’ (Reference Dinges2015) linguistic evidence is coherent with a linguistically sensitive, externalist treatment of ‘know’ where disambiguation of the concept of knowledge employed is linguistically feasible (by means of doxastic attitude verbs and modifiers). Thus, the strongest version of ‘the argument from semantic awareness’ is coherent with SPI and can be resisted if SPI is shown to be independently explanatorily powerful. Of course, it is a different matter altogether whether SPI is independently explanatorily powerful.Footnote 38
