Feinberg's targets in this essay are the claim that people are only capable of pursuing their own self-interest, and its corollary that altruistic or selfless actions are not possible (except when they happen to be perceived means to one's own self-interest). These two claims constitute "psychological egoism.” Note that psychological egoism is a descriptive claim about people's motivations; as such, it does not prescribe any behavior. Thus, it would be wrong to conclude from psychological egoism that one ought to act in one's self-interest, for psychological egoism merely claims that people do act only in their own self-interest, regardless of whether doing so is good or bad.
Much of Feinberg's essay is devoted to refuting a number of popular arguments in favor of psychological egoism. One such argument is that because all of our actions originate in our own desires, it seems that our actions are always the result of pursuing our own ends or seeking to satisfy our own desires. One interpretation of this account of motivation is that all of our actions are selfish – the claim of psychological egoism. Feinberg points out, however, that this reasoning is faulty. The fact that all of our actions are prompted by our own desires is a tautology (i.e., it is trivially true), for all it means for an action to be one's own is that it is prompted by a motive that comes from oneself. But one cannot conclude from a tautology a factual claim such as "all of our actions are prompted by a particular kind of motive, viz., a selfish motive," which is the real claim of psychological egoism.
Feinberg's final critique of psychological egoism concerns questions about the status of the theory itself. That is, even if the egoist concedes that one cannot conclude, on the basis of the arguments Feinberg discusses, that all people act from selfish motives, the egoist might nonetheless argue that people's motives are in fact always selfish. To evaluate that claim, Feinberg distinguishes analytic statements, the truths of which are determined solely by the meanings of the words in the statements (e.g., "all bachelors are unmarried"), from synthetic statements, the truth of which are determined by observation (e.g., "all bachelors are neurotic"). If the psychological egoist's claim is synthetic, there must be some conceivable evidence that the egoist would recognize as counting against the claim that people always act selfishly. On the other hand, if the egoist insists that every conceivable example of human behavior is motivated by selfishness, then the claim is analytic. Consider the case of someone giving his last dime to a person in need. This would seem to be evidence against the claim that all human actions are selfish, and it should count against psychological egoism construed as a synthetic claim. If, however, the psychological egoist maintains that her definition of 'selfish' includes the person giving up his last dime, then she commits the fallacy of the suppressed correlative. That is, she has expanded the definition of 'selfish' so greatly that there is nothing that counts as unselfish to contrast it with. But 'selfish' is a word that requires a contrasting word in order to make it meaningful, and such an expansive notion precludes that contrasting word. Thus, defining 'selfish' so broadly that it encompasses all actions only fallaciously precludes the possibility unselfish actions.