> A bigger problem for psychological egoism is that some behavior does not seem to be explained by self-regarding desires. Say a soldier throws himself on a grenade to prevent others from being killed. It does not seem that the soldier is pursuing his perceived self-interest. It is plausible that, if asked, the soldier would have said that he threw himself on the grenade because he wanted to save the lives of others or because it was his duty. He would deny as ridiculous the claim that he acted in his self-interest.

> The psychological egoist might reply that the soldier is lying or self-deceived. Perhaps he threw himself on the grenade because he believed that he could not bear to live with himself afterwards if he did not do so. He has a better life, in terms of welfare, by avoiding years of guilt. The main problem here is that while this is a possible account of some cases, there is no reason to think it covers all cases. Another problem is that guilt may presuppose that the soldier has a non-self-regarding desire for doing what he takes to be right.

> The psychological egoist might reply that some such account must be right. After all, the soldier did what he most wanted to do, and so must have been pursuing his perceived self-interest. In one sense, this is true. If self-interest is identified with the satisfaction of all of one’s preferences, then all intentional action is self-interested (at least if intentional actions are always explained by citing preferences, as most believe). Psychological egoism turns out to be trivially true. This would not content defenders of psychological egoism, however. They intend an empirical theory that, like other such theories, it is at least possible to refute by observation.

> There is another way to show that the trivial version of psychological egoism is unsatisfactory. We ordinarily think there is a significant difference in selfishness between the soldier’s action and that of another soldier who, say, pushes someone onto the grenade to avoid being blown up himself. We think the former is acting unselfishly while the latter is acting selfishly. According to the trivial version of psychological egoism, both soldiers are equally selfish, since both are doing what they most desire.

> The psychological egoist might handle apparent cases of self-sacrifice, not by adopting the trivial version, but rather by claiming that facts about the self-interest of the agent explain all behavior. Perhaps as infants we have only self-regarding desires; we come to desire other things, such as doing our duty, by learning that these other things satisfy our self-regarding desires; in time, we pursue the other things for their own sakes.

> Even if this picture of development is true, however, it does not defend psychological egoism, since it admits that we sometimes ultimately aim at things other than our welfare. An account of the origins of our non-self-regarding desires does not show that they are really self-regarding. The soldier’s desire is to save others, not increase his own welfare, even if he would not have desired to save others unless saving others was, in the past, connected to increasing his welfare.

https://plato.stanford.edu/entries/egoism