Relativism and Moral Luck revisited

Contextualist and relativists regarding epistemic modals quarrel on how to account for phenomena like the following:

Sally’s mother comes into Sally’s bedroom to find her looking under the bed. “What is going on?” asks Sally’s mother, “why are you looking under the bed?” (1) “My glasses, they might be there,” replies Sally. After taking a long look under he bed, Sally finds no glasses under it. So she moves on to look in other places, but not before saying, (2) “Oops, I was wrong.”

What is going on here?

According to me, this is just a case of moral luck, but in the context of assertion.

According to Andrew Latus (2001):

The problem of moral luck traps us between an intuition and a fact:

. 1)  the intuition that luck must not make moral differences (e.g., that luck must not affect a person’s moral worth, that luck must not affect what a person is morally responsible for). 

. 2)  the fact [Berg-Cross 1975, Cushman et.al. 2009, Cushman 2008, etc.] that luck does seem to make moral differences (e.g., we blame the unfortunate driver more than the fortunate driver).

For the relativist, cases like Sally’s can be characterized by a similar tension, since it also seems to trap us between an intuition and a fact:

. 1)  the intuition that luck must not make semantic differences, for example that luck must not affect what a person is responsible for when he makes an assertion 

. 2)  the fact that luck does seem to make semantic differences (e.g., we apologize from unfortunate falsehood).

In my original paper, 2012, I asserted that responses to the problem of moral luck have been of two broad sorts. Some claim that the intuition is mistaken, that there is nothing wrong with luck making a moral difference. Others claim that we have our facts wrong, that luck never does make a moral difference. The first sort of response has been the least popular. (Latus 2001)

Just as in the case of moral luck, in my 2012 paper, I claimed that responses to the problem of what I called “semantic luck” – i.e., cases like Sally’s – had been also of two broad sorts. Some (relativists) claim that the intuition is mistaken, that there is nothing wrong with luck making a difference in truth conditions. Others (antirelativists) claim that we have our semantic facts wrong, that luck never makes a difference in an assertion’s truth conditions.

On Williams’s view, for example, (i) the truck driver is (i) innocent, (ii) unlucky, and (iii) justifiably feeling bad after the fact because (iv) he did something awful.

He is (i) innocent because the fatal accident cannot be traced back to any negligence on his part, yet he was unfortunate enough (ii) to run over the child. Furthermore, (iii) he is justified in feeling bad because (iv) he actually did something awful: he killed a child.

A William’s analogue for the Sally case could analogously say that Sally is (i) innocent (of asserting that the glasses might be under the bed), (ii) unlucky (that the glasses were not there), and (iii) justified in her post facto retraction because (iv) she actually said something false.

But now, 2015, I have realised that the menu of options is actually larger than I thought in 2012:

MENU OF OPTIONS

I. One act or (at least) two acts?

Case a. Multi-action
Dissolve the paradox by introducing a distinction between what Sally asserted and what she asserted might happen. The first one is true and that is why she was right in asserting it, the second is false, and that is what she recognises at the end.

II. One action + One norm or (at least) two norms? Accepting that what was asserted in (1) was also what was retracted in (2), leaves open the question of whether there is one or two norms regulating what we are justified and obliged to assert and/or retract.

Case b. Dualism!

There is no such thing as the norm of action.

There is no such thing as the norm of assertion either.

We are always evaluating assertions according to at least two criteria: is the person justified in asserting what she is asserting? and is what the person is asserting true? Most times, both criteria match in their outcome, so we do not notice. Cases like Sally’s are odd precisely because in them these two criteria diverge and we have conflicting intuitions.

For every value v, there is an associated value j (for justification) such that if it is appropriate to judge an action a as to whether it is v, then it is also appropriate to judge whether it is j, where j and v are associated in such a way that if an action is j but not v, then we are faced with a case of bad luck, and if the action is v but not because of its being j, then we are faced with a case of good luck.

Making an epistemic assertion like Sally’s is a lot like gambling (as a matter of fact, for a long time, the working title of my 2012 paper was “Semantic Gambling”) in so far as no matter how much information we have, most times we cannot rule out all possibilities that what we are asserting is false. We are always taking a risk, but some risks are rational (and thus we are justified in acting on them), while others are not; and even the smallest risk can not pay off.

Sally retracts, no because the risk was not rational (she was justified at t1), but because her gamble did not pay off (at t2).

Who would hold this? I think I would. Also, virtue epistemologists. 

Notice that the claim is that there are at least two criteria for evaluating assertions, but most likely there are more. It is very likely that we also evaluate assertions according to whether the asserter has knowledge of what she asserts, and thanks to Gettier, we know this cannot be reduced to the aforementioned two criteria.

III. One action + One norm + Contextualism or Relatiivism? The debate between contextualists and relativists presupposes that what was asserted in (1) was also what was retracted in (2), and that there is one single norm regulating what we are justified and obliged to assert and/or retract.

Case c. Relativism
Sally was right to say that the glasses might be under the table because by her epistemic base at t1, it was true; she was right to apologise at t2 because by her new epistemic base, it was false.

If what was asserted in (1) was also what was retracted in (2), it must have been something that Sally knew (at t1) before looking under the bad (something like (4) ’For all Sally knew at the time of her assertion), her glasses could have been under the bed’), but also something she could rightly retract (at t2)  after looking there (like (3) ‘Sally’s glasses were under the bed’ or (5) ‘For all Sally would come to know at the time of her retraction, her glasses were under the bed’). Presumably, it must have been something that was true at t1 (like (4)) but false at t2 (like (3) and (5)). There cannot be such a proposition within the limits of traditional contextualism. Therefore—relativists conclude—we require a new kind of proposition that is sensitive both to features of the setting or context of assertion (t1) and to the context of its assessment (t2).

Who holds this? MacFarlane, Egan

Notice that relativism and dualism are very similar: both try to take on board both the intuition that luck must not make semantic differences, and the fact that luck does seem to make semantic differences. But they do it differently. For the dualist, Sally’s assertion was right in one sense (that is, according to certain criteria) and wrong in another (that is, according to other different criteria); while the relativist thinks that Sally’s assertion was right and wrong under the same criteria (the single criteria that norms assertion and retraction), but in different contexts: right under the relatively poor informational base she had at t1, and wrong under the relatively richer informational base she had at t2.

IV. One action + One norm + Contextualism: Right or Wrong?
Accepting that what was asserted in (1) was also what was retracted in (2), that there is one single norm regulating what we are justified and obliged to assert and/or retract, and that propositions do not change truth value, what is right and what is wrong to assert does not depend on how much information one has at the context according to which (but not necessarily from which) the assertion is evaluated, still leaves the question open as to whether wat Sally said was right or wrong.

Case d. Negative Contextualism: 

Adopt an invariantist or contextualist interpretation of the modal operator and accept that Sally’s assertion at t1 is false. Sally’s use of “might” makes her assertion sensitive to a contextually determined epistemic standard, and we (both Sally and us) were wrong about what those standards were. Since epistemic standards are sensitive to what is stake, we did not realise that the stakes were higher, until the bad consequences became evident at t2.

What is the danger here? We might end up with a theory according to which it is never ok to say anything.
How to avoid the danger? There is empirical evidence that we absolve people from some very unforeseeable bad consequences.

Who would hold this? Bradbury

Case e. Positive Contextualism: 

Adopt an invariantist or contextualist interpretation of the modal operator and accept that Sally’s assertion at t2 is false. That is why it is correct to reply “It’s O.K., it wasn’t your fault,” “You could not have known,” or something else of the sort. Compliment with a nice psychological theory that frames the human tendency of blaming unlucky agents who actions have (relatively unforeseeable) negative consequences  as a sort of cognitive bias.

What is the danger here? We might end up with a theory according to which it is always ok to say anything.

Can you bite the bullet? Yes, maybe it is always ok to say anything you believe to be true.

Who would hold this? Nagel