The meaning of "selfishness"

People often talk of behaviour as being motivated by selfishness or self-interest. These terms feature in explanations of specific actions (“they did it for selfish reasons/out of self-interest”), in claims about someone’s character (“they are selfish”), and in more general claims about human behaviour (“human behaviour is at root selfishly motivated/motivated by self-interest”).* How are we to interpret them?

On the one hand, there is an as it were formal or academic definition - in terms of necessary and sufficient conditions - which says that an act or omission whose purpose is, wholly or in part, intended to benefit oneself is self-interested or selfishly motivated. Thus according to this definition, the fact that an act or an omission is selfishly motivated doesn’t automatically entail that it’s wrong or ethically problematic. For instance, this definition implies that trying to get enough food to stay alive is (normally) selfishly motivated. But we don’t find that ethically problematic.

But we often don’t think in terms of necessary and sufficient conditions when we are using concepts. Instead, we think in terms of prototypes: that an action is selfish to the extent to which it resembles prototypical cases of selfishness. And these prototypical cases of selfishness are arguably often ethically problematic. They include someone who takes more than their fair share of the benefits of some joint project, or someone who consumes luxury goods without considering the effects on the environment. More extremely, they might even include corruption and other sorts of crimes.

These two ways of thinking about selfishness are often conflated in ways that can be unhelpful. First, it can lead to an overly cynical view of human behaviour. Some people might identify selfishly motivated actions using the first definition. Since that definition is wide, they’ll find that most behaviour is selfishly motivated, in one way or another. But then in the next step, they conflate the wide definition and the prototype-based definition - e.g. because their associations with the term “selfishness” largely derive from the latter. That leads them to infer that much or most of this behaviour is at least somewhat similar to the prototypical cases of selfishness - which tend to be ethically problematic. And that leads them to conclude that human behaviour largely has dark motivations. But that doesn’t follow, since according to the wider definition of selfishness, selfishly motivated behaviour need not be ethically problematic.

But there is also a converse fallacy. In this case, people rather identify selfishly motivated actions using the prototype-based definition. That leads them to think that selfishly motivated behaviour is not overwhelmingly common. But then they conflate the two definitions again, and that leads them to think that most behaviour isn’t selfishly motivated even in the first, wider sense. And that may lead them to underestimate, e.g. the efficacy of incentives that speak to people’s selfish motivations (in the first sense of the word).

For instance, consider employees who don’t go out of their way to optimise a company’s operations. This behaviour is usually not seen as particularly ethically problematic, and is thus hardly a prototypical case of selfishness. Thus if you conflate the two senses of “selfishness”, you might infer that this kind of behaviour isn’t based on misaligned motivations but that it is rather due to, e.g. some cognitive error. Alternatively, you might fail to see it altogether. This might, in turn, cause you to fail to address it properly. In many cases, you could make employees do more to increase the company’s productivity through providing the right sorts of incentives - e.g. prizes or the chance of promotion. But if you don’t see these omissions as motivational problems (because they’re not “selfish” in the narrow sense), then you won’t take those steps.

In my view, selfish motivations in the wide sense are strong and ubiquitous - but they often work in subtle ways, and in ways which aren’t ethically problematic (or at least not strongly so). For instance, if employees are provided with the aforementioned kind of incentive, they might find their mind drifting towards problems at work much more frequently than it previously did. They might not necessarily have been irresponsibly neglectful previously, but these problems might have tended to slip their mind. Thanks to the incentives, they are now more focused. I think these kinds of subtle mechanisms causing people to be more alert when they benefit personally are extremely common, and that the aggregate effect is that incentives speaking to people’s self-interest, in the wide sense, make a major difference. But that doesn’t necessarily mean that these incentives will cause them to engage in an ethically problematic, prototypically selfish behaviour.

One upshot of this for effective altruists is, I think, that we should use incentives that speak to people’s self-interest to encourage higher-impact actions. Indeed, we already do that. The exact extent to which we should do it must be settled empirically (probably largely via trial and error), but there doesn’t seem to be any principled reason against using these kinds of incentives. Self-interested actions, in the wide sense, aren’t necessarily as problematic as one might think. Thus one shouldn’t necessarily see such incentive systems as a concession that humans are fundamentally bad, or anything like that. Instead, one can take a more relaxed approach towards sensible incentive systems.

*The terms “self-interest” and “selfishness” are slightly different - “self-interest” being more formal and less moralising - but I think that they are so close that associations about them influence each other. Therefore, I treat them as one package here.