In a brief email conversation, I suggested to Ned Markosian that a complete account of mercy should take into account the agent's intentions and beliefs. For example, I said, if a judge thought he could sentence a convict to five or ten years, and out of compassion, chose to sentence the prisoner to five years, then the judge was acting mercifully even if the judge was wrong about his options and was in fact legally required to sentence the convict to five years.
Markosian offers a counterexample to the idea that intentions and beliefs must figure into an account of mercy:
Tom mistakenly thinks that he has a right to paint Tim's bicycle blue (because Tim raised his eyebrows when Tom said "Hi" to him), and also mistakenly thinks that painting Tim's bicycle blue would result in more harm to Tim than not doing so. Then, out of compassion for Tim, Tom chooses not to paint the bike.
We're at a kind of impasse, because I think this is an example of a merciful action. Tom's action in refraining from painting Tim's bike was merciful. It was also, of course, completely mistaken in almost every important way (we are supposed to assume, I think, that Tom is wrong to think he has a right to paint the bicycle and that doing so would harm Tim). Tom's action was deficient in some epistemological sense. But as to mercy—well, the act is merciful. He (to his mind) could have done worse to Tim, and had compassion on Tim, and so refrained.
How can such an impasse be resolved?
Markosian's own account offers a suggestion here. One of it's strengths is that it offers an explanation as to why intuitions differ about mercy in some cases. On his account, the mercifulness of an action rests not just on the action and its context, but also on the context of the utterance of the sentence "this action was merciful." So for example, if the context of the utterance makes legal principles the most relevant, then a judge who chooses a lesser sentence might be merciful thereby, but if the context of the utterance makes moral principles the most relevant, then the lesser sentence may be neutral or even unmerciful, depending on the situation. (For example, if giving the lesser sentence was positively immoral, then the action would be ipso facto immoral, since on Markosian's account an impermissible action is never merciful by the lights of that permissibility standard).
Could it be in this case that Markosian and I are thinking of the situation as though from two different contexts, thereby accounting (on Markosian's account) for our differing judgments? This may be so, but it would be hard to figure out what those two contexts would be without a considerably more detailed conversation. As it is, it sure looks like a single context—Markosian and I, two philosophers (albeit he's much more schooled in this particular area of philosophy) talking about a single case in an email exchange we both know everything about.
Aggravatingly, a difference in contexts might be outlined as follows: I'm judging from the standpoint of someone who takes an agent's state of mind into account, whereas Markosian is judging from the standpoint of someone who doesn't. So for me, the permissibility standard I see as most relevant is permissibility according to the agent's own lights, while Markosian is presumably applying a permissibility standard which doesn't take the agent's own lights into account. (I call this "aggravating" for the petty reason that it would count in favor of Markosian's account).
We shouldn't mistake this for any kind of relativism about mercy. It's not relativism—it's contextualism. It's not that Markosian and I can both be right even though we disagree, it's that we're making our judgments in two different contexts, and different contexts make for different truth conditions. In truth, on a contextualist account, we don't disagree, we're simply making two different kinds of judgments under the same name.
Well, we do disagree about something—we disagree about what the appropriate permissibility standard is. My own gut feeling is that the agent's beliefs and intentions are always relevant. Markosian's is either that it's never applicable or applicable in only some contexts.
My claim that it's always relevant is a pretty strong one and I'm not at this particular moment sure exactly how to defend it. In fact one important consideration has occurred to me that counts against it. It's natural to suppose that the recipient of mercy owes the merciful agent thanks. But think of Tom and Tim in the example above. Does Tim owe Tom any thanks for Tom's refraining from painting the bicycle? This seems dubious! But if it's true that the recipient of mercy owes thanks to the merciful agent, then I will have to defend a claim that Tim should thank Tom!
To be honest, I do think that if Tim were appraised of Tom's state of mind, it could indeed be appropriate for Tim to thank Tom—perhaps after correcting his misconceptions, but still thanking him in the "I know you meant well!" sense—but "appropriate" is not the same as "obligatory."
But is it obligatory for the recipient to thank the merciful agent? Or is it only obligatory in certain types of cases. I'll think about it. What do you think? (How many wheels am I reinventing here? I need to hie me to the library tomorrow.)