widespread view (call it V) holds that regret for a forgone lesser good is rational only when that good is of a different kind than the greater good-for example, when it is an instance of knowledge rather than of pleasure. My article challenges V, arguing that rational regret requires the lesser good to be distinct from the greater only in the weak sense of being a different instance of some kind of good, and not in the strong sense of being an instance of a different kind of good. I consider three types of examples: where you chose a greater pleasure for person a over a lesser pleasure for person b, a greater pleasure at time t1 over a lesser pleasure at time t2, and a greater pleasure with one set of introspectable properties over a lesser pleasure with another set, e.g., a greater pleasure of eating bagels over a lesser pleasure of discussing philosophy. In all these examples, I argue, the forgone lesser pleasure is not included in the greater pleasure as a part and can therefore be rationally regretted even if what makes it good is just its being a pleasure.

In framing the issue as about "rational" regret I do not assume, as Johnson supposes, any substantive theory of rationality needing further explanation or even any specific view such as maximizing about what makes actions right. I take the term "rational" from the existing literature, especially Stocker, and use it to mean "intrinsically appropriate," which for me implies (though this is a further claim) "intrinsically good." The idea is that some attitudes to goods and evils are in themselves fitting, and the question is when regret is among them.

The article starts by deriving the rationality of regret from a more general "proportionality view" about the division of appropriate pro-attitudes. If greater good x and lesser good y both do not exist, you should desire x more intensely than y, but also y to some degree, namely one proportioned to its degree of goodness. If x and y do exist, you should take more pleasure in x but also some pleasure in y. The situation is similar if you have chosen x over y. Then you should feel more pleasure that x exists but also a proportioned regret or wish that y, which does not exist, did.

McIntyre challenges this derivation on the ground that it treats regret as just a "general pro-attitude" rather than one with importantly distinctive features. She claims first that regret is directed at states that are not only valued but also disvalued, since they are recognized as lesser goods. But surely regret for a forgone outcome is rational only in so far as the outcome has goodness, not in so far as it lacks it. She also claims that regretting is a sui generis attitude, different from deploring, resenting, and decrying in being essentially directed at objects of choice. I am not persuaded that regret is sui generis. But even if it were sui generis, this would not have to affect the tie between its rationality and proportional division. Desire for a non-existent good is a different mental state from pleasure in an existing one. Even so, we can ask how both the intensity of desire and the intensity of pleasure are best divided among goods, and can expect the answers to these parallel questions to likewise be parallel. Similarly, even if regret is sui generis, the question of its rationality can turn entirely on more abstract properties it shares with other pro-attitudes.

The connection between regret and other pro-attitudes is not tangential to my critique of V, but motivates it. Imagine that you could produce either a greater good x or a lesser good y. Is it necessary, for you to have some desire for y, that you think of it as a different kind of good from x? Surely not. Or imagine that x and y both exist. Is it necessary, for you to take some pleasure in y, that you have that thought? I claim that the case of regret is identical..

The article is not as clear as it should be about exactly what the object of rational regret is. The simplest possibility is that you regret the forgone good y considered on its own, wishing that it existed because on its own it is good. It may be objected that this regret is not the kind at issue in V, since it is not regret that y does not exist rather than x. But there is another possibility. Before you act, you can have some desire for the outcome in which y exists and x does not because that outcome would be good, even though not as good as the outcome with x and not y. In a similar way, after your choice you can feel some regret that the y-and-not-x outcome does not exist, even though you feel more pleasure that the x-and-not-y outcome does. Here the object of your regret does involve the existence of y rather than x, as required. It may be objected that the regret is still not of the right kind, since it involves only the thought that the forgone outcome would be good, and not the thought that it would be better than the chosen outcome. McIntyre insists on this "better than" thought in her review, as did Stocker in his book. In response to Stocker, my article says that if you have chosen a greater pleasure for a over a lesser pleasure for b, you can regret the forgone outcome because, though not better overall, it is better for b or in b's life. Even given a single value, you can find a lesser good in one respect better.

McIntyre challenges this response, saying it grounds the rationality of regret in "a kind of error," that of evaluating outcomes in light only of some relevant information, namely about b's life and not a's. This is an interesting argument, and may even be sound. (I am not sure.) But I do not see how it does not apply equally to pluralist explanations of regret. Imagine that you have chosen some more valuable pleasure over some less valuable knowledge. Pluralists say you can regret the forgone outcome because, though less good overall, it would have been better with respect to knowledge. Does this not involve a similar "error," namely that of evaluating outcomes in light only of some relevant values? If McIntyre's argument is sound, it precludes both monists and pluralists from grounding regret in "better than" thoughts. But this may not matter if regret can take complex objects of the y-and-not-x kind. And if McIntyre's argument is not sound, monists as well as pluralists can regret what is better only in some respect.

Stocker defends his previously published claim that although regret is rational in my three examples, that is because they really involve plural values, e.g., the pleasure-of-a as a different kind of good from the pleasure-of-b. (I call this "extreme pluralism.") His main argument is that how many values a theory recognizes depends not just on its claims about goodness but on its "function" from those to other moral claims.

The first part of the argument concerns claims about rightness. Stocker says that if a theory tells us not just to maximize pleasure but to care who gets it, it treats different people's pleasures as different kinds of goods. But this is too quick. My article argues that there are simpler explanations for a concern about distribution than Stocker's, e.g., that we value equality-in-pleasure as a second good in addition to pleasure. My argument here is brief, and the issue needs more discussion. But let me make two other points. Because an argument about rightness does not apply to my second two examples, where maximization remains, it cannot support all of V's claims about regret. And it is questionable whether it supports any claims about regret. Even if extreme pluralism were needed to explain our concern for distribution, why would it follow that it is also needed to explain rational regret, especially if, as this part of Stocker's argument allows, there can be rational regret for forgone instances of the same kind of good at different times or with different accompanying qualities?

The main part of Stocker's argument is therefore its second, which claims that a theory's allowing rational regret itself requires the theory to be pluralistic. In my article I say this kind of claim begs the question against my position and reduces V to an empty tautology, one saying only that a theory that allows regret must allow regret. Stocker resists the charge of begging the question and asks in turn how I justify my claim that a theory that allows regret can be monistic. I do so not on the basis of the (bad) arguments he suggests I may be motivated by, but as follows. A theory is monistic if it says there is a single, uniform explanation of the goodness of all good things. In doing so, I implicitly reject part of Stocker's claim about functions. I think how many values a theory recognizes depends ultimately only on its claims about values. We may test for those claims indirectly, by seeing which are needed to explain its claims about rightness, but we cannot read them directly off its other claims, as the second part of Stocker's argument does. So far as I can see, that second argument still begs the question in favour of V.

In looking to "functions," Stocker assumes that extreme pluralism is worth defending only if it has some other significant implications. I agree, arguing in Section 3 that if a simpler theory has all the same implications as Stocker's, we should prefer the simpler theory. I grant in principle Brook's point that the more complex theory may be true, in that its posited entities actually exist, and my claims in Section 3 should therefore be adjusted. But if a moderate and an extreme pluralism are otherwise equivalent, the choice between them, though perhaps still open, is not of much moral interest. It will be made, if it can be, not on moral grounds but by arguments from the general theory of universals. The philosophers who defend V, and extreme pluralism as a corollary of V, do so because they think them necessary to account for important features of our moral practice. If those features can be explained more simply, as I claim, the main motive for defending V vanishes.