It is a mistake to believe that someone is selfish just because they get pleasure from helping someone else- although it is understandable why this mistake has been made by psychological egoists.
There are several reasons that make their argument seem plausible, including:
i) every voluntary action is prompted by motives or desires or impulses which are the agent's own and nobody else's;
ii) when a person gets what they want they feel pleasure, and therefore people pursue other things only as a means to their own pleasure.
iii) people deceive themselves into thinking they want something noble when what they truly desire is to receive gratitude from others, to congratulate themselves, or to have a good conscience;
and iv) people are inclined to behave according to the 'sanctions of pleasure and pain'.
In response, to say that someone gets pleasure as a result of their actions doesn't imply that what they always want is their own pleasure - It may be that the person desired the good of another person, animal, or mankind in general, that they wer motivated to action.
The very fact that pleasure is experienced as a result of an action presupposes that the person desired somthing else.
There is also a paradox in assuming that all people are motivated to action by the sole desire fro their own pleasure, simply because the exclusive desire for pleasure is self-defeating: only if you really want or value other things as ends in themselves will you feel pleasure from obtaining them.
Although it may be true that the real motivation behind all apparent altruism is selfish, there is simply not enough empirical evidence to either prove or disprove this claim.
As for moral education, there is a limit to what the 'sanctions of pleasure and pain' can teach. If the anticipation of a reward or punishment was the only guide for behaviour, there would be no reason why people shouldn't do something immoral if they can get away with it. However, it seems there are other reasons for behaving morally, therefore this theory is not sufficient for explaining all human action.
Lastly, there is a difference between self-interested and selfish behaviour. The Egoist's definition of 'selfish' has been extended in meaning to include 'self-interested', that is, for the egoist, 'selfish' doesn't merely mean "regarding one's own comfort, advantage, etc. in disregard of, or at the expense of that of others," instead it has taken on a new meaning synonymous with "acting from one's own motives", thereby committing the fallacy of the suppressed correlative - since 'selfish voluntary action' means 'motivated voluntary action,' according to this definition, there can be no 'unselfish,' that is to say, 'unmotivated' voluntary actions at all.
Even if we accepted the Egoist's new definition, we would eventually distinguish between two types of 'selfish actions' : those which regard the interests of others and those which don't, and that only the latter are reprehensible.
The egoist must prove that every voluntary action is prompted by a selfish motive, not just state that every voluntary action is voluntary (self-motivated), as this proves nothing.
Therefore, in the case which is stated in the question above, it is clear that the action is not selfish, because it disregards one's own comfort, etc. for the benefit of someone else.