UPDATE 1: It’s probably worth reading this article by Feinberg, which explains why self-interest ≠ whatever you happen to care about. See also the “Further Update” below before making a comment.

Peter Singer made famous a thought experiment like this:

You are walking down the street when you see a small toddler drowning in a shallow pool. You can save the toddler easily, but only if you jump in right now. Doing so will destroy your hard-earned smart phone, costing you $500.

Most people judge that we must save the child here; it would be wrong not to do so.

Now, what does ethical egoism say about this case? Remember, there might be a large gap between what Rand thinks egoism implies about this case versus what egoism actually implies.

Egoistic theories are a proper subset of consequentialist moral theories. Egoistic moral theories try to explain the rightness and wrongness of actions entirely in terms of how they affect the agent’s self-interest. There are more and less stringent version of egoism. Consider:

1. Maximizing egoism: An action is right for an agent if and only if it contributes more, on net, to that agent’s self-interest than any other available action.
2. An action is right for an agent if and only if it contributes, on net, to that agent’s self-interest.
3. An action is right for an agent if and only if doesn’t harm, on net, the agent’s self-interest.

1 is more demanding than 2, which is more demanding than 3. 1 forbids you from doing anything other than action that best promotes your self-interest. 2 allows you to do anything that promotes your self-interest; you needn’t do the best thing. 3 simply forbids you from harming your self-interest, but allows you to perform actions that have zero marginal utility.

1-3 are specified in terms of the the action’s actual consequences, but you can generate 3 new versions of egoism by replacing “contributes” with “is expected to contribute”. So, for instance, a maximizing egoist might instead hold that an act is right for an agent if and only if that actions is expected to contribute more, on net, to the agent’s self-interest than any other available actions.

Which version does Rand endorse? Well, half the time she seems like she’s not an ethical egoist at all, but rather a neo-Aristotelain eudaimonist with a formal egoist meta-ethics. But the other half she seems to me to vacillate between 2 and 3. I don’t know whether Rand is an actual or expected utility-egoist, so I’ll just gloss over this point here.

Now, consider what a egoist would have to say about the drowning toddler example. Each form of egoism implies that you must let the toddler drown. (An expected utility egoist might say you can save the toddler, but only if you have a sufficiently high chance of getting a sufficiently high reward for saving the toddler). Even the weakest form of egoism says not merely that you may let the toddler drown, but rather that it would be wrong to save the toddler. After all, it’s specified in the thought experiment that saving the kid imposes a cost upon you.

Randians might balk that these kinds of thought experiments are unrealistic. But they’re just plain wrong about that. Actuality implies possibility. I’ve been in a drowning toddler-type case, though that the unaccompanied toddler was about to walk into traffic rather than drown. I did not receive any reward for saving the toddler, by the way, and it cost me a good hour’s worth of my time.

FURTHER UPDATE:

Many people (including Rand) are deeply confused about what counts as egoistic action. That’s one reason why I recommended that everyone read the Feinberg piece up top.

Consider the following argument: Saving the kid is egoistic, because I care about kids. After all , whenever a person acts voluntarily, she acts to promote something that she is concerned about. If a person acts to promote something she is concerned about, she is acting according to one of her interest. If a person acts to promote one of her interests, then she is acting out of an interest of the self. If she is acting out of an interest of the self, then she is acting out of self-interest. Therefore, if I voluntarily save the kid, I save him out of self-interest.

The bolded premise is false. This premise conflates “interest of a self” with “self-interest”. It confuses the issue of who cares (who has the concern) with the issue of what is cared about (what the object of concern is). The fact that something is a concern of mine doesn’t imply that I am actually concerned with myself when I act on behalf of that thing. If I care about the drowning toddler and thus save him, I am acting to promote his welfare, not my own. It’s true that I am interested in his interests, or that his interests are an interest of mine, but that’s just a funky way of saying that I am altruistic. To be altruistic is to hold other people’s welfare is an end in itself.

If you’re a normal person, you’re not a sociopath. You sympathize with others. You feel bad when you see them suffer, and you feel happy when you see them happy.

Does that mean that when I promote someone’s else’s welfare, I’m really just trying to produce pleasure sensations in myself? Or that when I save the toddler, I’m really not concerned with the toddler, but rather view saving him merely as instrumental to avoiding bad feelings in myself?

No. If my sons died, I’d be sad. But the reason I feel joy when things go well for them and sad when things go badly is that I love them for their own sake–I view them as ends in themselves apart from my own welfare. Consider: Suppose my younger son is hurt. A genie appears and gives me two options. 1. He fixes my son’s injury. 2. He casts a spell instantly killing my son, erasing him from everyone’s memory, erasing all traces of him, and thus allowing us to go on as if he never existed at all. If I were just trying to avoid the bad feelings, I’d be indifferent between these two options. But I’m not–I’d pick option 1 over option 2, hands down. This means that I’m concerned not merely to avoid bad feelings, but to help for his sake. Again, it means I’m genuinely altruistic.

Also, ask why do we feel guilt when we fail to help drowning toddlers? Sociopaths don’t care about other people for their own sake–they view them merely as instruments. Sociopaths don’t feel guilt for failing to help others. The reason you feel guilt when you fail to help is because you care, or because you think you have a duty to help apart from your self-interest. If you genuinely didn’t care, if you thought that the drowning toddler had no more intrinsic value than a tumbleweed about to fall over a cliff, you wouldn’t feel guilt when you let him drown. You’d feel nothing at all.