Normative Uncertainty and the Dependence Problem

Link post

Pod­gorski, Abe­lard (2020). Nor­ma­tive Uncer­tainty and the Depen­dence Prob­lem. Mind 129 (513):43-70.


In this pa­per, I en­ter the de­bate be­tween those who hold that our nor­ma­tive un­cer­tainty mat­ters for what we ought to do, and those who hold that only our de­scrip­tive un­cer­tainty mat­ters. I ar­gue that ex­ist­ing views in both camps have un­ac­cept­able im­pli­ca­tions in cases where our de­scrip­tive be­liefs de­pend on our nor­ma­tive be­liefs. I go on to pro­pose a fix which is available only to those who hold that nor­ma­tive un­cer­tainty mat­ters, ul­ti­mately leav­ing the challenge as a threat to re­cent skep­ti­cism about such views.

I was un­aware there was skep­ti­cism about nor­ma­tive un­cer­tainty, speci­fi­cally the idea that it’s only de­scrip­tive un­cer­tainty, not nor­ma­tive un­cer­tainty, that cre­ates un­cer­tainty about what is best to do.

A key quote from the in­tro­duc­tion that mo­ti­vates the pa­per and ex­plains this is­sue more:

A num­ber of the­o­rists (Lock­hart 2000, Ross 2006, Sepielli 2009, MacAskill and Ord 2018, Tarsney 2018) have tried to provide ac­counts of de­ci­sion-mak­ing un­der nor­ma­tive un­cer­tainty. But there has been a back­lash against this en­tire pro­ject from a con­tin­gent of philoso­phers who ar­gue that only your de­scrip­tive un­cer­tainty mat­ters for what you ought to do (Weather­son 2014, Har­man 2015, Hed­den 2016).
In this pa­per, I draw at­ten­tion to a puz­zling class of cases which the con­tro­versy over nor­ma­tive un­cer­tainty has ne­glected – cases in which our cre­dences about de­scrip­tive facts de­pend on our cre­dences about nor­ma­tive facts. I will show that most ex­ist­ing views about de­ci­sion mak­ing, both those that take nor­ma­tive un­cer­tainty into ac­count and those that do not, give un­ac­cept­able recom­men­da­tions in these cases. But the prob­lem, I’ll go on to ar­gue, is much worse for views that don’t take nor­ma­tive un­cer­tainty into ac­count at all. To avoid the coun­terex­am­ples, views must be sen­si­tive not only to both kinds of un­cer­tainty, but also to the re­la­tion­ship be­tween them. As I’ll show, there is a rel­a­tively painless strat­egy for in­cor­po­rat­ing this kind of sen­si­tivity into ex­ist­ing views that re­spect nor­ma­tive un­cer­tainty, but no similarly easy solu­tion in reach for those who claim that only de­scrip­tive un­cer­tainty mat­ters. Ul­ti­mately, then, the prob­lem is am­mu­ni­tion against the re­cent scep­ti­cism of the­o­riz­ing about nor­ma­tive un­cer­tainty. In­deed, it raises doubts that there are any in­ter­est­ing norms that are sen­si­tive to merely de­scrip­tive un­cer­tainty.

From there things get fairly tech­ni­cal, and I don’t think I can offer a good sum­mary, but the pa­per con­cludes thusly:

But I have tried to give rea­son to hold out hope that a defence is pos­si­ble. De­ci­sion-mak­ing un­der nor­ma­tive un­cer­tainty is a re­search pro­gram that is still in its in­fancy, and it would be a mis­take to give up on it too soon, if there seem to be pow­er­ful rea­sons in its favour. At the same time, the ar­gu­ment sug­gests a note of cau­tion to those de­vel­op­ing an ac­count of nor­ma­tive un­cer­tainty: many prin­ci­ples at­trac­tive at first blush will fall apart if we do not at­tend to the ways our nor­ma­tive and de­scrip­tive be­liefs are in­te­grated.

This gives the im­pres­sion that ideas about nor­ma­tive un­cer­tainty are start­ing to ma­ture and gain at­ten­tion among a wider philo­soph­i­cal au­di­ence.