I also don’t agree with the claim that rationality does not tell us anything about what we should want. Perhaps instrumental rationality doesn’t, but epistemic rationality does.
This is potentially misleading. First, there’s a good sense in which, moral realism or no moral realism, instrumental rationality does play a strong role in telling us what we ‘should’ want—far more than does epistemic rationality. After all, instrumental rationality is often a matter of selecting which lower-level desires satisfy our higher-level ones, or selecting which desires in general form a coherent whole that is attainable from the present desire set.
But by ‘what we should want’ you don’t mean ‘what we should want in light of our other values’; you seem rather to mean ‘what we should want in light of objective, unconditional moral facts’. (See my response to EY.) You’re right that if there are such facts, then insofar as we can come to know them, epistemic rationality will direct us toward knowing them. But you haven’t defended the independent assumption that knowing moral facts forces any rational agent to obey those facts.
Let’s assume that moral realism (in the sense you seem to mean—what I call moral unconditionalism) is true and, moreover, that the relevant facts are knowable. (Those are both very big assumptions, but I’m curious as to what follows.) How could we then argue that these facts are internalist, in the strong sense that they would be completely ineffable to any rational agent who was not thereby motivated to obey the dictates of these facts? In particular, how can we demonstrate this fact non-trivially, i.e., without building ‘follows the instructions of any discovered moral facts’ into our definition of ‘rational’?
Does our concept of epistemic rationality, in itself, require that if agent A learns normative fact N (say, ‘murder is wrong’), A must then eschew murder or else be guilty of e-irrationality? Clearly not. E-rationality is only about making your beliefs better fit the world. Further actions concerning those beliefs—like following any instructions they contain, or putting them in alphabetical order, or learning the same facts in as many different languages as possible—are extraneous to e-rationality.
(ETA: A perfectly e-rational agent is not even required to follow normative facts purely about beliefs that she learns, except insofar as following those norm-facts happens to foreseeably promote belief-accuracy. A purely e-rational agent who learns that murder is objectively wrong will not thereby be motivated to avoid murdering people, unless learning that murdering is wrong somehow leads the agent to conclude that murdering people will tend to make the agent acquire false beliefs or ignore true ones.)
Does our concept of instrumental rationality, in itself, require that if agent B learns normative fact N, B must then eschew murder or else be guilty of i-irrationality? Again, it’s hard to see why. An agent is i-rational iff it tends to actualize situations it values. If B’s values don’t already include norms like ‘follow any instructions you find embedded in moral facts’, then there seems to be no inconsistency or (i-)irrationality in B’s decision to disregard these facts in conduct, even if the agent is also completely e-rational and knows with certainty about these moral facts.
So in what sense is it (non-trivially) irrational to learn a moral fact, and then refuse to follow its dictate? What concept of rationality do you have in mind, and why should we care about the relevant concept?
This is potentially misleading. First, there’s a good sense in which, moral realism or no moral realism, instrumental rationality does play a strong role in telling us what we ‘should’ want—far more than does epistemic rationality. After all, instrumental rationality is often a matter of selecting which lower-level desires satisfy our higher-level ones, or selecting which desires in general form a coherent whole that is attainable from the present desire set.
But by ‘what we should want’ you don’t mean ‘what we should want in light of our other values’; you seem rather to mean ‘what we should want in light of objective, unconditional moral facts’. (See my response to EY.) You’re right that if there are such facts, then insofar as we can come to know them, epistemic rationality will direct us toward knowing them. But you haven’t defended the independent assumption that knowing moral facts forces any rational agent to obey those facts.
Let’s assume that moral realism (in the sense you seem to mean—what I call moral unconditionalism) is true and, moreover, that the relevant facts are knowable. (Those are both very big assumptions, but I’m curious as to what follows.) How could we then argue that these facts are internalist, in the strong sense that they would be completely ineffable to any rational agent who was not thereby motivated to obey the dictates of these facts? In particular, how can we demonstrate this fact non-trivially, i.e., without building ‘follows the instructions of any discovered moral facts’ into our definition of ‘rational’?
Does our concept of epistemic rationality, in itself, require that if agent A learns normative fact N (say, ‘murder is wrong’), A must then eschew murder or else be guilty of e-irrationality? Clearly not. E-rationality is only about making your beliefs better fit the world. Further actions concerning those beliefs—like following any instructions they contain, or putting them in alphabetical order, or learning the same facts in as many different languages as possible—are extraneous to e-rationality.
(ETA: A perfectly e-rational agent is not even required to follow normative facts purely about beliefs that she learns, except insofar as following those norm-facts happens to foreseeably promote belief-accuracy. A purely e-rational agent who learns that murder is objectively wrong will not thereby be motivated to avoid murdering people, unless learning that murdering is wrong somehow leads the agent to conclude that murdering people will tend to make the agent acquire false beliefs or ignore true ones.)
Does our concept of instrumental rationality, in itself, require that if agent B learns normative fact N, B must then eschew murder or else be guilty of i-irrationality? Again, it’s hard to see why. An agent is i-rational iff it tends to actualize situations it values. If B’s values don’t already include norms like ‘follow any instructions you find embedded in moral facts’, then there seems to be no inconsistency or (i-)irrationality in B’s decision to disregard these facts in conduct, even if the agent is also completely e-rational and knows with certainty about these moral facts.
So in what sense is it (non-trivially) irrational to learn a moral fact, and then refuse to follow its dictate? What concept of rationality do you have in mind, and why should we care about the relevant concept?