Knowledge rule of Assertion

This section shows Williamson’s argument for the thesis Knowledge Rule of Assertion, i.e., one must: assert p only if one knows p. Before showing Williamson’s argument, I would like to discuss the status of this rule, which is crucial to this argument.

Human has speech acts, say, query, greeting, warning, promise, command, conjecture, assertion, to mention a few. Every speech act is governed by norm. Most speech acts are governed by more than one norms. For example, query is governed by be polite, be explicit, be relevant, be sincere. As a speech act, assertion is also governed by many norms, say, be true, be informative, be relevant, be sincere, be warranted, be well phrased, be polite. The knowledge rule, i.e., one must: assert p only if one knows p, is just one norm among these norms of assertion.

Incidentally, some norms are shared by different speech acts. For example, the norm be explicit is shared by speech acts query, command, and assertion. However, to characterize a speech act, one needs not to show every norm of it, but just needs to show the norm(s) distinguishing it from other speech acts. Just like specifying rule(s) of an examination E, we never try to identify all rules of E. I seldom see any examination whose rule includes ‘participant should not cheat’. This rule may need to be specified only if we were specifying the rules of examination. When we are specifying the rules of E, however, we need not specify the rule ‘participant should not cheat’

because E is already an examination. In order to characterize E, we only need to specify those rules which are not shared by other examinations. To tell who Henry V is, one never needs to tell every characters of him, but only the character distinguishing him from others, say, he is The King of England who is on his throne from 1413 to 1422. Assertion shares norms with other speech acts;

thus to characterize assertion, we only need to specify those norm(s) which distinguishes assertion from other speech acts. Williamson calls the norm(s) distinguishing assertion from other speech acts constitutive rule(s) of assertion. If assertion is a speech act, it should have constitutive rule, which distinguishes it from other speech acts. Williamson presupposes that assertion is a speech act so that it has constitutive rule. For the sake of convenience, ‘rule’ refers to constitutive rule hereafter.

Constitutive rule has an important characteristic. That is C is constitutive rule of speech act A only if C necessarily govern A. Since C distinguishes A from other speech act, C govern A in every situation. In this sense, C necessarily governs A. Asking why C is governing A doesn’t make any sense; just like asking why the rule participant must not cheat governs examination doesn’t make any.

Other than presupposing that assertion has constitutive rule, Williamson also presupposes that assertion only has one rule. Obviously, the simpler a theory is, the better it is. In construction a theory of assertion, the fewer rules the theory involves, the simpler it is. If we can find one workable rule to characterize the speech art assertion, this characterization should be the best.

Under these two presuppositions, Williamson argues for the thesis Knowledge Rule of Assertion which claims: one must: assert p only if one knows p.

Truth rule is the first candidate for the rule of assertion. Truth rule claims: one must: assert p only if p is true. However, we use truth to appraise speech acts other than assertion. For example, conjecture, guessing. A true conjecture is better than a false conjecture. A true guessing is better than a false one. Truth is a norm of the speech acts of conjecture and guessing as well as assertion.

Constitutive rule of assertion only governs assertion. Therefore truth rule could not be constitutive rule of assertion. This simple argument shows that truth rule is not constitutive rule of assertion.

Although truth rule is not constitutive rule of assertion, truth is still a norm of assertion. Other than assertion, truth is also a norm of conjecture, guess, and swear. Guess, conjecture, assertion, and swear all share another norm, i.e., evidential norm. However, the evidential standard of these speech acts are different from each others. This difference suggests that we can distinguish assert from other speech acts by its unique evidential standard. The following story shows that assertion has an extremely high evidential standard, i.e., one must: assert p only if the evidential probability of p is 1, i.e., the probability of p is 1 based on one’s evidence. Consider the following story.

‘Suppose that you have bought a ticket in a very large lottery. Only one ticket wins. Although the draw has been held, the result has not yet been announced. In fact, your ticket did not win, but I have no insider information to that effect. On the merely probabilistic grounds that your ticket was only one of very many, I assert to you flat-out ‘Your ticket did not win’, without telling you my grounds.’ (Timothy Williamson, 2000, p.246)

Let p be ‘Your ticket did not win’. p is true. Also, I believed p. Since p is highly probable on my evidence, I am justified in believing it. Although I am justified truly in believing that p, I am not yet warranted to assert it. This story shows that justified truly believing p is not enough to be warranted to assert p. Further more, it shows that assertion requires an extremely high evidential probability

standard; I am not warrant to assert p unless the probability of p base on one’s evidence is 1.

Observe that no matter how low is the winning probability of your ticket, I am still not warranted to assert p. This means that one is warranted to assert p only if p has probability of 1 based on one’s evidence.

One may worry about that this consequence might induce sceptic consequence, since the probabilities of most our ordinary assertions based on one’s evidence are seemingly lower than probability 1; however, this would happen only if one takes that evidential probability of p is 1 means that one is certain with p. However, one should not take the former as the latter. For Williamson, that one is certain with p means that one would not giving p no matter what would happen. In previous section, I mentioned that one may lose one’s evidence. Evidential probability is based on one’s evidence, thus p which has probability of 1 based on one’s evidence now may drops to below 1 later. Consider Williamson’s example. ‘I toss a coin, see it land heads, put it back in my pocket and fall asleep; once I wake up I have forgotten how it landed.’ (Timothy Williamson, 2000, p.219) Obviously, before I fell asleep, that the coin lands head is included in my evidence;

therefore, according to EV, the evidential probability of that the coin lands head based on my evidence is 1 since the proposition is included in my evidence. After I fell asleep, that the coin lands head is no longer included in my evidence, since I have forgotten how it landed. According to EV, the evidential probability of that the coin lands head is 1/2 based on my evidence since the proposition is no longer included in my evidence. This example shows that one may lose evidence because one forgot, however, one may lose evidence by gaining new evidence as well. The example of drawing black balls in the previous section shows one lost evidence even one doesn’t forget any relevant information. When I put the red ball in the bag, I saw the red ball; thus I had the evidence that there is a red ball in the bag. The evidential probability of that there is a red ball in the bag based on my evidence is 1 at that time. However, after 10000 drawings of black ball, I wonder whether I mis-remember or mis-seen that is a red ball. I no longer believe that there is a red ball in

the bag, according to E = K, the proposition is no longer included in my evidence, since one knows p entails one believes p. The evidential probability of the proposition drops below 1. One should not confuse that p having probability of 1 based on one’s evidence with that one is certain with p, since one may lose one’s evidence so that probability of p may drop below 1.

One is warranted to assert p only if the probability of p based on one’s evidence is 1. How to explain this phenomena? Williamson’s explanation is that one must: assert p only if p is included in one’s evidence. According to EV, If e is included in one’s evidence, then the evidential probability of e is 1. Although EV does not entail that p is included in one’s evidence whenever the probability of p based on one’s evidence is 1, it seems to me that there is no good reason to reject Williamson’s explanation. Because E = K, that one must: assert p only if p is included in one’s evidence just means that one must: assert p only if one knows p. That is the thesis Knowledge Rule of Assertion.

One may challenge that the thesis Knowledge Rule of Assertion is counter-intuitive. There are some situations wherein the knowledge rule of assertion isn’t answered while the assertion is reasonable. For example, knowing that it is the last second for catching your train, I assert ‘that is your train’ without knowing that. Intuitively, my assertion is reasonable since it is the last second to catch your train. I am reasonable to assert that even I don’t know that is your train. However, one should not confuse that one’s assertion is reasonable with that one’s assertion is permissible.

Although my assertion is reasonable, it is not permissible. Sometimes, real life forces one to break rule in order to fulfill greater good. For example, the old story telling that in order to save a life one is forced to lie in some situations. Of course, lying in such situation is reasonable, still one broke the rule of assertion so that one’s assertion is not permissible. One is sometimes reasonable in lying doesn’t entail that knowledge rule is not in force. Sometimes different rules might conflict with each other. In such situations, some rule overrides others. I assert that is your train without knowing that is your train. My assertion is reasonable because it is the last second you catch your train. The knowledge rule is overrode. Obviously, if the situation was not that bad, you could resent that I

don’t know that. You didn’t resent because the knowledge rule is overrode; but that doesn’t mean I did not break the rule of assertion. Real life is a mess, sometimes it forces us to do impermissible.

4.3 Criticism

In this section, I will examine criticism of Hindrik who claims that assertion is just expressing one’s belief.

In the previous section, I indicate that we should not confuse that an assertion is reasonable with that an assertion is permissible. Frank Hindriks confused these two concepts. Hindriks claims that the traditional analysis of assertion (linguistic expression of belief) provides an excellent start point for arguing that assertion is indeed governed by a knowledge rule. He says,

As it turns out, then the traditional analysis of assertion as the linguistic expression of belief (BE), provides an excellent point of departure for defending the idea that assertion is indeed governed by a knowledge rule.

(Frank Hindriks, 2007, p.405)

For Hindriks, asserting p in itself is just expressing a belief p but one needs not believe in p.

(BE) To assert that p is to utter a sentence that means that p and thereby express the belief that p.

(Frank Hindriks, 2007, p.400)

Notice that BE is not a normative claim. It seems to me that, for Hindriks, assertion in itself has not any norm governing it. However, if one assert p when one does not believe p, still one is breaking a norm. If assertion in itself has not any norm governing it, what norm did one break then? Hindriks acknowledges that in normal situation, one must: asserts p only if one believes p. For Hindriks, this is just a consequence of another norm, that is

(NS) In situation of normal trust, one ought to be sincere. (Frank Hindriks, 2007, p.401)

For Hindriks, that one sincerely asserts p means that one believes p. When (NS) applies to assertion, we have:

(NSAB) In situations of normal trust, one must: express the belief that p only if one believes that p.

(Frank Hindriks, 2007, p.402)

For Hindriks, the rule that one must: express the belief that p only if one believes that p is in force only in situations of normal trust. In other words, the norm is coming from situations of normal trust, not from assertion itself. Hindriks accepts Williamson’s claim

(RBK) One must: believe that p only if one knows that p. (Frank Hindriks, 2007, p.403) From (NSAB) and (RBK), we have

(RAK*) In situations of normal trust, one must: assert p only if one knows that p. (Frank Hindriks, 2007, p.403)

From RAK*, Hindriks claims that one must: assert only if one knows that p in situations of normal trust. That means, RAK* does not necessarily govern assertion. Both NS and RAK* are not constitutive rule of assertion.

It seems to me that the motive of Hindriks taking this stance is that there are some situations in which one is permissible to lie.

Imagine, for instance, a Nazi asking you whether there are Jews in your house. If you are in fact hiding Jews because you want to protect them from being deported, we deem it permissible to lie to the Nazi. (Frank Hindriks, 2007, p.402)

I cannot find any reason in Hindriks’s paper to support that in the situation it is permissible to lie to the Nazi. Of course, we deem that in such situation, it is reasonable to lie to the Nazi, but that would not entail that it is a permissible lie.

Hindriks claims that knowledge rule is not constitutive rule of assertion, it only governs some assertion, but not all of them (Frank Hindriks, 2007, p.393) Obviously, NS is neither constitutive rule of assertion since it governs only situations in normal trust. Nor BE is constitutive rule of

assertion since it is totally not a rule at all. Hindriks never suggests another constitutive rule for assertion. Could assertion have not any constitutive rule?

If knowledge is able to be communicated, then the tool we use to communicate should be assertion. If assertion is the tool for communicating knowledge, then it must have constitutive rule.

Although Williamson himself have not argued for assertion has constitutive rule, it seems to me the foregoing reasoning supports the claim that assertion has constitutive rule. Hindriks does not, however, provide any reason to claim that assertion has not constitutive rule.

4.4 Conclusion

I have demonstrated Williamson’s argument for the thesis Knowledge Rule of Assertion, i.e., one must: assert p only if one knows p. Williamson shows that the evidential standard to assert p is extremely high: one must: assert p only if the probability of p based on one’s evidence is 1. By EV, i.e, e is evidence for h for S if and only if S’s evidence includes e, P (h⎮e) > h, Williamson interprets that one must: assert p only if the evidential probability of p as that one must: assert p only if p is included in one’s evidence. Since E = K, we have one must: assert p only if one knows p.

By the thesis Primeness, the obtaining of the condition that one knows p hinges on the case which one is in. Since one must: assert p only if one knows p, the obtaining of the condition that p is assertible should also hinges on the case which one is in. The case which one is in dominates assertibility of p. In Chapter 2, I showed Williamson’s argument for the thesis Anti-KK Principle, which claims that there is a case wherein one knows p but one does not know that one knows p.

Since the obtaining of the condition that p is assertible is hinges on the case which one is in, there should be case wherein p is assertible but that p is assertible is not itself assertible. In fact, Williamson have provided an argument there is such a case. Consider the following navigator’s story.

Imagine an early navigator sailing unknown seas on a slowly moving boat.

He wonders whether there is land ahead (at any distance: assume for simplicity that he does not know that the earth is round). Early in the morning, he has no idea; it is clear to him that no land is yet visible.

Gradually something appears on the horizon. At first he is not sure whether he is imagining it; even after he is sure, he has at first no idea whether it is land or a mere bank of clouds. The former hypothesis slowly gains in probability over the latter. After several hours there is no doubt. By evening the boat is moored to land. The navigator is phlegmatic; his confidence that there is land ahead grows as slowly as the visible scene changes; he experiences no flash of conviction. The whole process is gradual.²¹(Williamson, 1995, pp7-8)

Let t0 be a time early in the morning, tn be the time the boat is moored to land; and for each i, ti be one second interval between t0 and tn; also for each i, Ci be the case corresponding to each ti; P be

‘there is land ahead’. By the similar argument in chapter 2, we have

(1i) If in Ci it is assertible that it is assertible that P, then in Ci - 1 it is assertible thatP.

Since the boat is moored to land in Cn, P is assertible in Cn. (2n) P is assertible in Cn.

For the sake of reductio ad absurdum, we assume

(3i) if P is assertible in Ci then it is assertible that P is assertible.

By similar process of the argument in Chapter 2, we got a absurd consequence (20) P is assertible in C0.

Instances of (1i) is true, and (2n) is also true by the story; hence, some instance(s) of (3i) fail. There is a case wherein p is assertible but that p is assertible is not assertible.

21 Williamson, T. 1995. ‘Does Assertibility Satisfy The S4 Axiom?’. CRITICA, Vol. XXVLL, No. 81, diciembre 1995:

Chapter 5 Conclusion

Toward an Elegant Model for Knowledge, Evidence, and Assertibility

In chapter 4, I showed Williamson’s argument for the thesis knowledge rule of assertion, i.e., one must: assert p only if one knows p. In ordinary discourse, we do make assertions to communicate with others; and we can understand each other in most cases. Now, if Williamson’s argument holds, knowledge is communicated by assertion. Consequently, we have an answer to the question if knowledge is communicable, how? We communicate knowledge by assertion.

In fact, the knowledge rule of assertion is essentially nothing more than a theoretical consequence of the thesis that knowing is a mental state. To see this, let us review all theses I have demonstrated in previous chapters. They includes

(i) the thesis that knowing is a mental state (KMS);

(ii) the thesis that the condition that one knows p is prime (the primeness thesis);

(iii) the thesis Safety Requirement of Knowledge;

(iv) the thesis Anti-Luminosity;

(v) the thesis Anti-KK Principle;

(vi) the thesis Anti-Luminosity of Evidence;

(vii) the thesis E = K;

(viii) the thesis Knowledge Rule of Assertion.

In the previous chapters, I illustrated each argument for thesis (ii) to (viii). I deliberately emphasized the presupposition of each argument for the theses. The emphasis on Williamson’s presupposition had two aims: firstly, to indicate that each thesis is well-grounded. Secondly, to reveal that the theses (ii) to (viii) are all theoretical consequences of thesis (i). In this chapter, I will summarize the arguments we have seen in the previous chapters. In the summary, the reader can clearly see that each thesis in the list is just a theoretical consequence of the thesis that knowing is a mental state. Further, the reader will clearly see that the theses in the list provides us an elegant

In the previous chapters, I illustrated each argument for thesis (ii) to (viii). I deliberately emphasized the presupposition of each argument for the theses. The emphasis on Williamson’s presupposition had two aims: firstly, to indicate that each thesis is well-grounded. Secondly, to reveal that the theses (ii) to (viii) are all theoretical consequences of thesis (i). In this chapter, I will summarize the arguments we have seen in the previous chapters. In the summary, the reader can clearly see that each thesis in the list is just a theoretical consequence of the thesis that knowing is a mental state. Further, the reader will clearly see that the theses in the list provides us an elegant