In my post, Impugning the Desire-as-Belief Thesis, I argued that the Desire-as-Belief Thesis--namely, the claim that desires are a type of belief--is unacceptable as a way of unpacking the Broadly Anscombean View--namely, the claim that desires aim at the good. But even if one rejects the claim that desires are beliefs, one may still think that desiring to φ is always accompanied by the belief that φ is good. Let us refer to this proposal as the Desire-plus-Belief Thesis; the claim that the desire to φ is always paired with the belief that φ is good. The Desire-plus-Belief Thesis is offered as an alternative way of unpacking the Broadly Anscombean View, and should be distinguished from the weaker claim that an agent could only desire to φ if she has certain beliefs about φ. For example, one may think that one could only desire to φ if one also believed that it was possible for one to φ. However, such a proposal would not be a candidate for unpacking the claim that the good is the object of desire, for believing that one could φ may have little or no bearing on whether or not one believes that φ is good. Hence, the thesis that the desire to φ is always accompanied by some belief about φ is not the same as the Desire-plus-Belief Thesis; the latter is specifically concerned with the belief that φ is good.
An unqualified version of the Desire-plus-Belief Thesis seems highly implausible. For example, one may desire to have a cigarette even though one does not believe that it is good to do so. However, even if we find the preceding unqualified version of the Desire-plus-Belief Thesis implausible, we may still be tempted to buy into a qualified version of the thesis, according to which the desire to φ is always accompanied by the belief that φ is good from a certain perspective. For example, even though one does not believe that sleeping with someone who is not one's spouse is good simpliciter, one may still believe that it is good from the perspective of having one's sexual needs satisfied. Hence, according to the qualified version of the Desire-plus-Belief Thesis, an agent can only desire to φ if she believes that it would be good to φ from a certain perspective (even if she does not believe that φing would be good, simpliciter).
One objection to the Desire-plus-Belief Thesis (in both its qualified and unqualified forms) is that there seems to be putative cases of an agent desiring to φ, where the desire does not seem to be accompanied by the belief that φ is good from any perspective at all. Davidson's description of a man who has a “yen to drink a can of paint” even though he does not believe that it would be “worth doing so”, seems to be one such example. One very natural way of unpacking Davidson's claim that the paint drinker sees no “worth” in drinking the paint would be to say that the paint drinker does not believe that drinking the paint is good from any perspective whatsoever.
However, the preceding objection is certainly not the final word on the matter. There are some theorists, such as Warren Quinn and Thomas Scanlon, who would deny that the yen to drink a can of paint qualifies as a desire, in the relevant sense. To this end, Scanlon  distinguishes between a desire and a mere urge. According to Scanlon, the desire to φ is always accompanied by a tendency to see φing as good or desirable. Absent such a tendency, we have, not a desire, but an urge. Moreover, Quinn  notes that in cases like the paint-drinker example, the motivating state fails to rationalise the action or make it intelligible. That is to say, the yen of the paint drinker provides no justification for his actions, nor does it bring us any closer to understanding why the paint drinker did what he did. The upshot, according to Quinn, is that the yen of the paint drinker should not be considered a desire.
Although it is true that the yen of Davidson's paint drinker fails to rationalise or justify his actions, it is not clear (pace Quinn) that playing this rationalising or justificatory role is a necessary condition for a psychological state to be considered a desire.4 There is an alternative conception of desires, according to which, the presence of a desire serves the theoretical function of marking the difference between actions that are intentional—i.e., ones for which an agent is rationally or morally responsible—and actions that are not. Moreover, there is reason to think that the actions of Davidson's paint drinker are intentional by the lights of this alternative conception. For one thing, we may wish to hold that the paint drinker is rationally or morally responsible for his actions. For example, let us suppose that drinking the paint would result in his death, and that the paint drinker is aware of this fact. Let us suppose further that it is morally wrong to knowingly take one's life. Given that the paint drinker is aware that drinking the paint is fatal, we may very well wish to hold that he is morally responsible for taking his own life; to wit, there is nothing about Davidson's description of the painter drinker that would lead us to think that he should be excluded from such responsibility. Assuming that one is morally responsible for φing only if one φs intentionally, and given the view that desiring to φ is what distinguishes between cases in which one φs intentionally, and cases in which one does not φ intentionally, then it follows from the claim that the the paint drinker is morally responsible for drinking the can of paint that his yen to drink the can of paint is a desire. Consequently, if we see desire as playing the theoretical role of distinguishing between those actions that are intentional (understood as action for which an agent is rationally or morally responsible) and those actions that are not, then we may wish to hold that the yen of Davidson's paint drinker may be a desire.
I do not wish to settle the question of which of the two competing accounts just adumbrated is preferrable. The preceding discussion is simply meant to highlight that whether or not one regards the yen of Davidson's paint-drinker as a desire will depend on what one takes the theoretical role of desire to be. Since I do not wish to rule out the alternative account, according to which the yen of Davidson's paint-drinker may count as a desire, I believe we should leave room in our theorising for the possibility of such desires; namely, ones in which the desiring agent does not believe that the desired action is good from any perspective whatsoever. In short, the Desire-plus-Belief Thesis introduces a prejudice against certain accounts of the theoretical role of desire; a prejudice that I believe we do well to avoid at the present stage in our inquiry.
A more serious difficulty with the Desire-plus-Belief Thesis (in both its qualified and unqualified forms) is that it seems too weak to serve as a viable candidate for unpacking the Broadly Anscombean View; namely, the claim that the good is the object of desire. Let us assume, for the sake of argument, that desiring to φ is always accompanied by the belief that φ is good. Given this view, it seems more accurate to say that the good is the object of the accompanying belief than that it is the object of the desire itself. However, we have identified the Broadly Anscombean View with the claim that the good is the object of desire, not the claim that the good is the object of a particular kind of belief—namely, one that typically accompanies a desire. The problem is that (unlike the Desire-as-Belief Thesis) the Desire-plus-Belief Thesis preserves the desire's identity as a separate psychological state from that of the belief that accompanies it. Moreover, as far as the Broadly Anscombean View is concerned, it makes the good the object of the wrong psychological state; namely, the belief rather than the desire. Presumably, the defender of the Desire-plus-Belief Thesis would say that being accompanied by a belief that aims at the good is just what we mean when we say that a desire aims at the good. However, it seems odd to say that the desire to φ has the goodness of φ as its object simply because the belief that φ is good has the goodness of φ as its object. Why should the fact that the latter has the good as its object have any bearing on the former, given that they are distinct psychological states? In short, it remains unclear that the Desire-plus-Belief Thesis represents a genuine unpacking of the claim that the good is the object of desire.