One can do what is morally right without the action having moral worth, as with Kant’s (merely) prudent shopkeeper (not overcharging a naive customer when she easily could). The reason that the shopkeeper’s right action lacks moral worth is, roughly, that she does not have the right sort — the moral sort — of aim or motivation in doing the right (honest) thing.
In recent literature, folks distinguish two different sorts of motivational contents (motivating reasons) that might go into having the correct sort of motivation for a right action to have moral worth. First, there is content that concerns right-making features (RMF-type content), content like that overcharging this customer would fail to treat each customer equally. Second, there is content that concerns rightness (or wrongness) itself (RI-type content), content like that it would be wrong to charge this customer more than I charge the others. Some, like Nomy Arpaly and Julia Markovitz, have argued that only RMF-type motivational content is necessary. One intuitive point that works in favor of this approach is this: absent any knowledge of what makes a right action right, the insistence on doing the right thing can begin to look like a kind of empty fetish. Also, in the Huckleberry Finn case, Huck doesn’t turn Jim (the slave) in partly because he sees Jim as concretely human in various aspects (something that would make turning Jim in wrong, so an RMF-content) even though he thinks doing this is wrong (an RI-content that motivates toward turning Jim in).
More recently, others (e.g., Zoe Johnson King, Paulina Sliwa, Kashev Singh) have argued that RI-content is necessary as well (and sometimes that RI-content is primary, as Johnson King and Sliwa argue). Among other things, these authors (and others) produce reasonably convincing arguments against drawing the conclusions that Arpaly and Markovitz draw from cases of doing what is right simply because it is right (with no particular understanding of what makes right actions right operative, hence the charge of fetishism) and from the Huck Finn case (responding, by doing what is right, to the things that make the right action right, while holding explicit judgments that the right action is wrong).
I’m going to assume, without argument, that Johnson King, Sliwa and Singh are correct in thinking that RI-content — as well as RMF content — plays a necessary role in making a right action have moral worth. The intuitive idea here is something like this: in order for acting rightly to be the relevant sort of personal-level achievement or accomplishment (for moral worth to supervene) one must seek to do what is right (or avoid what is wrong) and exhibit adequate knowledge of what makes right (wrong) acts right (wrong). In the course of matters, it will be evident that (and why) I agree with Johnson King and Sliwa that RI-type motivational content is the more important or essential motivational content.
So consider this case: our functionally honest shopkeeper has a personal, idiosyncratic obsession (or fetish) with avoiding doing the wrong thing (and avoiding doing the more-wrong things when all options are wrong or presumptively wrong). Moreover, she has a definite idea of what is morally wrong (which happens to be the one that I favor): to do something morally wrong is to do something that it is normatively appropriate for affected others to resent or object to. And she has rich knowledge of what types of actions thus warrant resentment and why, including the type instantiated in this case (cheating). So she has the relevant RI and RMF content, but it does motivational work only because of her personal, idiosyncratic obsession or fetish with avoiding doing wrong. Call this sort of case the case of the fetishistically moral amoralist. I submit that this sort of functionally honest shopkeeper’s right action, in not cheating some given customer, does not have moral worth. Her motivation is not of the right sort. Yet it looks like it has the right sorts of motivational elements (motivationally efficacious contents that “push” in the right direction) for the going views to deem it morally worthy.
Plausibly, the problem here is that our avoiding-doing-wrong fetishist shopkeeper (super-reliably honest, better to do business with than the pretty moral shopkeeper down the street who slips up occasionally on his moral duties) does not have the right kind of intrinsic desire not to do wrong. But what does this even mean? Are there different kinds of intrinsic desires to avoid doing wrong? If so, what does the right sort of intrinsic desire (or aversion) look like?
Here is my proposal. Intuitively, avoiding doing wrong for distinctively moral motivations (or reasons) involves one experiencing guilt or shame upon doing (or even considering doing) wrong. But, clearly, the idea is not that one instrumentally desires to avoid doing wrong because one intrinsically desires not to experience guilt or shame. However, the following kind of broadly rational relationship can hold between distinct intrinsic desires: one intrinsically desires (or is intrinsically averse to) condition P (where this might be one performing a certain action or type of action) only because one intrinsically desires (or is intrinsically averse to) having some experience E that is reliably associated with P. Applying this model: one is intrinsically averse to doing wrong because one is intrinsically averse to the guilt or shame reliably associated with one doing wrong. This approach rightly treats the case of the fetishistically moral amoralist (the relevant functionally honest shopkeeper) as a case of right action that is not the sort of personal-level accomplishment that has moral worth (whatever, precisely, moral worth comes to — that is a separate question).
Note: because the focus here is on (the right kind of) intrinsic aversion to doing wrong, RI content (e.g., that this action is wrong) plays the leading content-driven motivational role. The role for RMF-type content is to provide adequate understanding about what makes right/wrong actions right/wrong. (Imagine someone who takes wrong action to be resentment-warranting action — but has no idea, not even implicitly, why various resentment-warranting types of action are resentment-warranting. Or, worse, in addition to this, she gets the types radically wrong, taking watching the sun rise to warrant resentment from her first-born son (and likewise for all mothers and sons) or some such lunacy.) So my analysis fits well with the priority that Johnson King and Sliwa give to RI-type content. But it also identifies the motivational element that explains this — a motivational element ignored if one addresses only the motivating elements of RMF-type and RI-type motivating contents or reasons. And that — and how — we intrinsically desire to avoid doing wrong seems to be what matters most or what does the most explanatory work.
(Further note. As I’m using ‘wrong’ in a “thin” sense that does not include wrong things being normatively impermissible for the agent (or generally), the sort of RI-type content that I’m speaking to here does not include the first-personally normative content that corresponds to using ‘wrong’ in a thick sense that references both wrongness (either under this generic description or under another, as with resentment-warrant) and first-personally normative features of action (like it being normatively impermissible to perform the action). And similarly for ‘right’. So there is potentially a distinct sort of RI-content-centered view of the required motivational elements that I am not addressing. I think we need to distinguish contents like that this action would warrant resentment (or the de dicto right/wrong content that this action is wrong) from contents like that this action is impermissible for me because it is resentment-warranting (and their potential motivating roles) — that is why I use ‘wrong’ and ‘right’ in the “thin” semi-descriptive or not-first-personally-normative sense. I have my doubts that, in order for a right action to have moral worth, the agent must represent her action as permissible for her, as the only option that is not impermissible for her, as what she has sufficient moral reason to do, or anything like that (most explicitly contra Singh’s approach). Perhaps such first-personally normative belief is necessary to carry off sophisticated sorts of deliberation.)