Here are some competitors in trying to formulate a norm of assertion:
The Knowledge Account: assert only what you know to be true.
The Defensibility Account: assert only what you can defend if challenged.
The Justified Belief Account: assert only what you justifiably believe.
In setting up a case to motivate his relativistic semantics for epistemic modals, Andy Egan here says the following:
To a first approximation, it’s appropriate to say that some utterance is true iff it’s true, and appropriate to say that an utterance is false iff it’s false. At the very least, the falsity of an utterance is an extremely salient candidate to explain the appropriateness of saying that it’s false, and the truth of an utterance is an extremely salient candidate to explain the appropriateness of saying that it’s true.
At first glance, this claim may not appear to be a competitor of the above accounts. But such appearances are deceiving, I think. To see why, let me give a little background. Egan is considering two assessments of the utterance ‘Bond might be in Zurich,’ one by those taken in by misleading evidence planted by Bond that he is in Zurich and the other by a present companion of Bond at headquarters in London. So the first remarks that the utterance is true and the second that it is false, and the assertions, one of which ascribes truth and the other falsity to the utterance, are appropriate assertions. We want a theory that accounts for this appropriateness.
In response to this desire, Egan proposes what I have quoted: that the first assertion is appropriate iff it is true and the second assertion is appropriate iff it is true. And now I’m tempted to make a philosopher’s fallacy, and attribute to Egan the following generalization of this proposal:
The Truth Account: assert what, and only what, is true.
But, after complaining about the fallacy, far be it from me to commit it! (Would that all fallacies were avoided so easily…) So I won’t commit it. Still, though, I wonder what, other than this generalization would make Egan’s setup warranted, and I can’t think of one. Maybe someone else can, though…
Suppose, however, that it is The Truth Account that lies behind Egan’s remarks. If so, there are serious problems for the view. First, norms of assertion ought to be more circumspect, not giving sufficient conditions for acceptable assertion but only necessary ones. As all of us know, there are lots of truths that are better left unsaid. The more important issue, however, is the other direction, and here the account faces serious obstacles, many of which are chronicled in Williamson’s work. Take a fair coin, and find two people taking different bets on whether it will land heads when flipped. If both are willing to assert what they are betting on, only one of them can be criticized appropriately on the basis of the norm of assertion.
One might think this example relies too much on future contingents having truth value, but I hold that it relies on this point to exactly the right extent! But we can change examples for the discomfited: let the bet be on the number of jellybeans in a particular jar. The general point is that vocalized guesses are criticizable on grounds intrinsic to the kind of action displayed: namely, that of assertion. It is useful to note that all three of the other accounts of assertion agree on this point: verbalized guesses violate the constitutive rules governing the practice of assertion.
Since I favor the Justified Belief Account, I won’t stop here before noting its fecundity with respect to Egan’s example. Note that only the Defensibility and Justified Belief Accounts can explain the appropriateness of both assertions that prompt Egan’s relativistic semantics. Since I’ve already given reasons to prefer the Justified Belief Account to the Defensibility Account here, primarily on grounds of operationalizing the justification component of the Justified Belief Account, I’ll not say anything more here about it. But if faced with choosing between the Justified Belief Account and a relativistic semantics for epistemic modals, I’d prefer the more conservative route…