1.

Philosophers often critically discuss whether psychological egoism (PE) and ought-implies-can (OIC), suitably combined with plausible further premises, support ethical egoism, the view that everyone morally ought always to do what best serves her own interest.Footnote ¹ These discussions nearly always include an assumption, variously worded or left implicit, the gist of which is this: PE, combined with OIC, entails OIB (below), a possible stepping-stone to ethical egoism.

In this paper I argue that PE and OIC, interpreted fairly, use “can” in different ways, a result of which, discussed in section 2, is that those theses do not entail OIB.Footnote ⁵ They entail something similar to OIB, but not OIB itself. From these facts further significant results follow, discussed in section 3. They challenge familiar arguments or assumptions about ethical and psychological egoism, and at least two of them pertain also to rational egoism, the view that everyone rationally ought always to do what best serves her own interest.

2.

Suppose that according to the evidence available to a person – to Tai, say – act A is clearly in his interest. However, owing solely to careless error or inattention, easily avoidable and still easily correctable, Tai lacks the belief that A is in his interest. With the possible exception of that belief (and its inevitable accompaniments, if any), Tai has everything he needs in the way of beliefs, desires, psychological traits, and so forth to choose to do A. Also, nothing else prevents Tai either from choosing to do A or from successfully carrying out that choice.

Given that Tai lacks the belief that A is in his interest, PE implies that Tai cannot do A. Many philosophers, I suspect, will object to this implication. They will do so even if they assume (perhaps arguendo, perhaps not) that a belief that A is in one's interest is causally required for an intention to do A. They will contend that even on that assumption, Tai can do A – and not just in an odd sense of “can.” Tai, they will say, has the power, the ability, to do A; the only thing holding him up is a momentary, avoidable, easily correctable lapse or error. It is thus implausible to say that he cannot do A.

Whether conclusive or not, this objection is too plausible to be brushed off. A truly charitable reading of PE will forestall it. On that reading, PE uses “can” in a particular sense. (At least, let us charitably assume a sense of “can” of the sort in question.) Using that sense, to say that a person can do an act is to say, at least in part, that he has all the psychological states which, owing to human nature, are causally required for an intention to do the act. A key point of PE is that those states include the agent's belief that the act is in his interest. Tai lacks that belief about A; thus, if PE is true, then even if Tai can do A in some sense, PE's implication that Tai cannot do A is true. For its meaning is that A is not an act Tai can do in the sense of “can” in PE.

Let us use “can_p” for the sense of “can” just described. PE, properly understood, asserts that a person can_p do an act, and thus has every psychological state required for an intention to do it, only if his psychological states include a belief that the act is in his interest.

Now, how should we read “can” in OIC? To read it as “can_p” is to invite an objection, namely, that on such a reading OIC is too permissive. It is too permissive in that it implies, implausibly, that a person might have no moral requirement to do an act – to heed the traffic lights on a busy highway, sayFootnote ⁶ – owing solely to an easily avoidable, easily correctable lapse of attention. For that is all it takes, sometimes, to lack one of the beliefs required for an intention to do the act (e.g., to brake for the lights, or to slow down). The general point is that moral requirements have a firmer hold on us than OIC would imply if OIC used “can” to mean “can_p.”

A possible response to this objection is this: No, it is not implausible that in that moment of easily avoidable, easily correctable inattention, during which the driver cannot_p heed the traffic lights (i.e., in which it is false that he can_p heed them), he has no moral requirement to heed the lights. He indeed has no such requirement. What he does have, and what can mislead us here, is a moral requirement to avoid, when possible, losing attention while driving on busy highways.

The objectors to whom this response is addressed will likely reject it, replying thus: We often have easily avoidable, easily correctable lapses of attention that are morally innocent. Why, then, are we morally required to avoid such lapses while driving on busy highways? The best answer is that, first, we have further, more basic moral requirements related to such driving, such as the requirements to avoid excessive speed, to stay in our own lane, and to heed the traffic lights. Second, those requirements remain in place during our lapse of attention, during which we cannot_p heed the lights. (Perhaps they would not remain in place if to say that a person cannot_p do an act implied that there is no robust sense of “can” in which the person can do the act. But it does not imply that.) That lapse, in other words, does not release us from those other moral requirements; rather, our obligation to avoid the lapse derives, at least partly, from those other requirements. An upshot is that if OIC is true, it does not use “can” to mean “can_p.”

The objectors' reply is plausible; so too is their original objection. Indeed, that objection, like the earlier objection to PE, is too plausible to be brushed off. A truly charitable reading of OIC will forestall it. Note that as OIC is generally, and plausibly, understood, its sense of “can” is such that, using that sense, to say that a person can do an act is not merely to say that his performance of the act is possible in a minimal way – for instance, that it is logically or metaphysically possible. It is to say that he can do the act in a robust sense of “can,” a sense implying that the act is somehow within his power, that the act available to him, an option for him.Footnote ⁷ However, in light of the above objection, we should not read “available,” “option,” and the like so that an act is available to a person only if he can_p do it – that is, only if he presently has every belief (desire, etc.) required for an intention to do it. Perhaps, as in Tai's case, one of those beliefs is easily attainable but not quite in place.

In sum, to be charitable to OIC we should read “can” in that thesis such that, to say that a person can do an act is to say that the act is among his options, his alternatives, his available acts, though not necessarily an act he can_p do. Let us use “can_o” for this sense.

So as I read OIC, it says that a person morally ought to do an act only if he can_o do it. Let me say three things about this reading.

First, as already evident, “can_o” does not imply “can_p.” More fully, the claim that Tai can_o do an act does not imply that he can_p do it.

Second, not only do I stand by my reading of OIC, but even aside from that fact, I consider my results significant. For I suspect that a great many philosophers accept my reading, at least tacitly. This is partly because my reading carries only a little information about the referents of “available acts” (mainly, that the agent can do them in a robust sense of “can,” though not necessarily in the sense denoted by “can_p”); thus, it accommodates a variety of views about the nature of those referents and about the exact meaning of OIC.

Finally, for PE and OIC to entail OIB, OIC must imply that a person morally ought to do an act only if he can do it, where “can” here means the same as it does in PE. However, OIC implies this only if its sense of “can,” “can_o,” implies “can_p,” the sense of “can” in PE. Thus, since “can_o” does not imply “can_p,” PE and OIC do not entail OIB.

They do, however, entail something similar to OIB. This is largely because, as I will show presently, PE, the view that a person can_p do an act only if he sees the act as in his interest, implies the thesis below. (Either that, or it implies something similar enough to that thesis that for my purposes, the thesis below will do.) I call it “IPE,” for “implication of PE.”

My claim is not that this thesis is true, but simply that PE implies it. My reason for this claim is this: To deny IPE is basically to say that sometimes, an act is available to a person, an option for him, even though, first, he lacks a belief that the act is in his interest, and second, no cogent reasoning and alertness to relevant facts could lead him to form that belief. This, it seems, is just a way of asserting, contrary to PE, that a belief on the person's part that the act is in his interest is not required for his intention to do the act.Footnote ⁹ In short, it seems that to deny IPE is to deny PE – or put differently, that PE implies IPE.

The conjunction of IPE with OIC entails the thesis below. Thus, the conjunction of PE with OIC does so as well.

To sum up thus far: Charity requires that we read “can” as “can_p” in PE; as “can_o” in OIC. Also, the claim that Tai can_o do an act does not imply that he can_p do it – in short, “can_o” does not imply “can_p.” A result is that contrary to a common assumption, PE and OIC do not entail OIB. They do, however, entail OIB*.

3.

The results of the preceding sections lead to at least five further results, each relevant to our thinking about ethical, psychological, or rational egoism. The final three are perhaps the most significant, but all five deserve mention.

First, to recall section 1, the question whether psychological egoism and ought-implies-can support ethical egoism receives frequent discussion, especially in ethics texts. Important in those discussions is the assumption that psychological egoism and ought-implies-can entail OIB. Hence, typically, such discussions pursue a red herring: the issue of how to supplement OIB so that it might (if true) support ethical egoism. The proper issue is how to supplement OIB* so that it might do that. (Is supplementing OIB* this way more challenging than doing so for OIB? I will attempt no answer to that question here – an adequate one would take us far afield. My point for now is that we cannot answer the question by pursuing the wrong issue.)

Second, a prevalent view is that psychological egoism clashes sharply with the content of morality, the word “morality” meaning either morality as ordinarily conceived or the prescriptions, in the real world, of the most common normative moral theories.Footnote ¹⁰ (A variation of this view is that psychological egoism severely limits what can serve as a plausible moral standard.)Footnote ¹¹ For if, as many think, OIC is true, and if PE and OIC imply OIB, then to grant psychological egoism is to imply that a great many acts that morality seems to require are not truly obligatory. They are not obligatory because the relevant agents do not see them as in their interest. Perhaps the agents are mistaken, or even irrational, in not seeing the acts this way, but that does not affect the point.

However, given that PE and OIC imply OIB* rather than OIB, then even with OIC granted, the clash between psychological egoism and morality is perhaps not so severe. Arguably, it is not severe enough either to be a powerful challenge to psychological egoism or, if psychological egoism is true, to necessitate revisions to morality that make it unrecognizable. For even if many of the acts that morality prescribes are not seen by the agents to be in their interest, it may be that for many of those acts, the human situation, as well as our own particular situation, is such that most of us would be reasonable to believe that the acts are in our interest. This is especially true if we take a broad view of what it is to be in our interest.Footnote ¹²

Third, a frequent claim (implication, or suggestion) is that psychological egoism, far from supporting ethical egoism, rules out ethical egoism.Footnote ¹³ However, the usual arguments for this claim, which are closely alike, run afoul of the distinction between “can_o” and “can_p.”

The following argument is representative: Ethical egoism, treated as a guide to life, entails that when acting, we morally ought to limit ourselves to acts that we see as in our interest. It thus implies that we can do otherwise than limit ourselves that way – that is, that we can do acts that we do not see as in our interest. This is because the claim that we morally ought to do X implies that we can do otherwise than X. If we cannot do otherwise than X, it is silly to say that we ought to do X. Thus, ethical egoism, treated as a guide to life, implies something – that we can do acts that we do not see as in our interest – that conflicts with psychological egoism.

This argument is open to multiple objections, but I will present just one. But first let me repeat that the argument is representative. It admits of various wordings, at least one of which – the substitution of “presupposes” for each instance of “implies” – affects its meaning slightly.Footnote ¹⁴ However, none of these rewordings materially affects my objection to the argument. Each necessitates, if anything, only minor changes.

My objection to the argument is this: Its key premise is that an implication holds between “We morally ought to do X” and “We can do otherwise than X.” However, this premise is plausible only if the sense of “can” it employs – the sense of “can” in “We can do otherwise than X” – does not imply “can_p.”Footnote ¹⁵ In fact, the premise is plausible only if the sense of “can” it employs is “can_o.” Note that this key premise is a sibling of OIC, namely, ought-implies-can-do-otherwise. Also, it is implausible to think that in this sibling, “can” has a different sense than it has in OIC, in which “can” means “can_o.” Thus, to the extent that the argument is correct in saying that ethical egoism implies, via its key premise, that we can do otherwise than limit ourselves to acts that we see as in our interest – that is, that we can do acts that we do not see as in our interest – the word “can” in that implication means “can_o.”

But then the implication does not conflict with psychological egoism, contrary to the argument's conclusion. Psychological egoism explicitly asserts that we can_p do an act only if we see it as in our interest, and implicitly asserts (via IPE) that we can_o do an act only if we either see it as in our interest or would be reasonable to do so. Neither of these assertions conflicts with the implication that we can_o do acts that we do not see as in our interest.

My fourth result concerns another tempting defense of the claim that psychological egoism rules out ethical egoism.Footnote ¹⁶ We can state that defense thus:

1. 1. Assumed premise: psychological egoism is true, and hence a person – call her Pia – can do an act only if she sees that act as in her interest.

2. 2. A situation, S, may exist in which, in terms of serving Pia's self-interest, act X is the best of her available acts; however, Pia does not see X as in her interest.

3. 3. OIC is true, and thus Pia morally ought to do act X in situation S only if she can do that act in S.

4. 4. Thus (from 1 and 2), it is false that Pia can do X in S.

5. 5. Thus (from 3 and 4), it is false that Pia morally ought to do X in S.

6. 6. Thus (from 2 and 5), ethical egoism is false: A situation may exist in which, in terms of serving Pia's self-interest, X is the best of her available acts (making it the act ethical egoism prescribes); yet it is false that in that situation, Pia morally ought to do X.

7. 7. Thus (from 1 through 6, via conditional proof), if psychological egoism is true, ethical egoism is false.

Those unaware of the results of section 2 might suspect that the first premise in the above argument, which says that Pia can do an act only if she sees that act as in her interest, conflicts with an element of the second premise, namely, that an act might be available to Pia even though she does not see the act as in her interest. But there is no such conflict. The element of premise 2 just mentioned basically says that sometimes, Pia can_o do an act even though she does not see the act as in her interest. It is thus consistent with the first premise, the meaning of which is that Pia can_p do an act only if she sees the act as in her interest.

Even so, the argument fails. The third premise, which asserts ought-implies-can, uses “can” to mean “can_o”; however, the fourth premise, being an implication of the first premise (psychological egoism) and the final clause of the second (“Pia does not see X as in her interest”), uses “can” to mean “can_p.” Since “can_o” does not imply “can_p,” the third and fourth premises do not entail the fifth.

Note that we cannot salvage the argument by revising the second part of premise 2 to say this: “however, Pia neither believes that X is in her interest nor would be reasonable to do so.” This revision may be tempting because it would make the first two premises entail a variation of premise 4 – namely, “It's false that Pia can_o do X in S” – which, combined with 3, entails 5. However, that variation of 4 conflicts with an element of the second premise, namely, that act X is available to Pia in S.

My fifth result relates to the previous ones. Specifically, at least two of the previous results, the third and the fourth, pertain not just to ethical egoism but also to rational egoism, the view that everyone rationally ought always to do what best serves her own interest.Footnote ¹⁷ This is because the two arguments those results concern, each for the claim that psychological egoism rules out ethical egoism, require only slight revisions to become arguments for the claim that psychological egoism rules out rational egoism. (In fact, the first of the two arguments is sometimes explicitly cast as an argument for the latter claim.)Footnote ¹⁸ However, the revised arguments have essentially the same flaws as the unrevised arguments. Let me demonstrate these points with respect to the second argument; that they apply equally to the first is not hard to see.

Consider the following:

1. 1. Assumed premise: psychological egoism is true, and hence a person – call her Pia – can do an act only if she sees that act as in her interest.

2. 2. A situation, S, may exist in which, in terms of serving Pia's self-interest, act X is the best of her available acts; however, Pia does not see X as in her interest.

3. 3*. OIC, tailored to the rational “ought,” is true, and thus Pia rationally ought to do act X in situation S only if she can do that act in S.

4. 4. Thus (from 1 and 2), it is false that Pia can do X in S.

5. 5*. Thus (from 3* and 4), it is false that Pia rationally ought to do X in S.

6. 6*. Thus (from 2 and 5*), rational egoism is false: A situation may exist in which, in terms of serving Pia's self-interest, X is the best of her available acts (making it the act rational egoism prescribes); yet it is false that in that situation, Pia rationally ought to do X.

7. 7*. Thus (from 1 through 6*, via conditional proof), if psychological egoism is true, rational egoism is false.

This argument fails in basically the same way as the argument of which it is a variant. This is partly because, with only slight revisions, the reasons for reading “can” as “can_o” in ought-implies-can apply also to “can” in the variant of ought-implies-can concerning the rational “ought.” (For instance, it may be that Pia rationally ought to do an act even if she cannot_p do it. For perhaps the only reason she cannot_p do it is that she lacks a particular belief – a belief she rationally ought to have.) Thus, premise 3* uses “can” to mean “can_o.” However, as said earlier, in premise 4 “can” means “can_p.” Since “can_o” does not imply “can_p,” 3* and 4 do not entail 5*.

4.

In conclusion, to read psychological egoism and ought-implies-can charitably is to read “can” as “can_p” in the first of those theses; as “can_o” in the second. Also, the claim that a person can_o do an act does not imply that the person can_p do it. A result is that although psychological egoism and ought-implies-can together entail OIB*, they do not entail OIB. Additionally, we have good reason to read “can” as “can_o” not just in ought-implies-can, but also in the variant of ought-implies-can concerning the rational “ought.” I think that if these facts were widely recognized, they would improve discussions of ethical, psychological, and rational egoism. For one thing, as shown above, they would prevent some common errors, each bearing on the import, viability, or possible combination of some of those forms of egoism.