Is rational self-interest tautological?

Comrade PhysioProf claims that rational self-interest as an explanatory paradigm for human behavior is tautological:

At some level, no matter what people do it must by definition by in their self-interest, or else they wouldn't do it. I tend to see this as a reductio ad absurdum argument for the uselessness of attempting to explain human behavior in terms of individual's "interests".

I think Comrade PhysioProf is mistaken in this view.

We could, of course, formulate a truly tautological view of rational self-interest: simply label "whatever motivates people to act" as "rational self-interest". In much the same sense we could formulate a truly tautological view of evolution and physics. We could label "whatever causes changes to organisms over time" as "evolution". But that we can do so does not mean that we must necessarily do so.

We can't infer much from a label in isolation from what the label applies to. (Thoughts makes the same fallacy in a different direction by inferring specific incorrect philosophical ideas from the adoption of other elements of the label "materialist"). We have to look at the details of what the label actually applies to. If we want to look at the label "evolution", we have to look at the details of actual evolutionary theory as proposed by actual scientists.

It's impossible also to actually falsify a true hypothesis, theory or paradigm. Because evolution as a paradigm is on the whole actually true, it's unlikely that we'll actually observe anything that would falsify the paradigm (as opposed to falsifying detailed hypotheses within the paradigm). What's important, then is seeing whether a paradigm could in principle be falsified.

What is the theory of rational self-interest? It's the theory that people have ideas in their head about what's better or worse in terms of their own subjective experience. They imagine, plan and anticipate the consequences of various possible actions in terms of the quality of their subjective experience. They then choose the action among the imagined alternatives that they believe will produce the best subjective experience.

One of the things that distinguises a scientific hypothesis, theory and paradigm from a tautology or definition is that the antecedent and the consequent can be independently verified. Consider the canonical syllogism:

P1: All people are mortal
P2: Socrates is a person
C: Socrates is mortal

If we can determine that Socrates is a person and independently determine that Socrates is mortal, then P1, which can be rephrased as the condition "if X is a person then X is mortal", becomes a scientific hypothesis. P1 would be a tautology if we could not independently determine mortality, either by not being able to determine mortality at all,

P1: All people have souls
P2: Socrates is a person
C: Socrates has a soul

or by explicitly predicating by definition the determination of "person" on the determination of "mortality":

P1: All people are mortal
P2: Lazarus is not mortal
C: Lazarus is not a person

If some statement is really a scientific theory, then it should be possible to create and verify a null hypothesis, an inverse conditional. (We have to make the above conditional more specific to make its inverse observable.)

Hypothesis: All people are mortal with a lifespan not exceeding 200 years/if X is a person, X will die within 200 years of his or her birth
Contrapositive: if it's not the case that all people are mortal, etc., then some person will not die after 200 years.

We could in principle observe a person living for 200 years. When we observe a lot of people, billions of them, we actually observe all of them dying well within 200 years. Science being what it is, our observations do not deductively prove the hypothesis, they strengthen our confidence in the hypothesis. It might be the case that the trillionth person we observe does not die; it might be the case that there's some other reason — perhaps they're all hiding from us — we do not observe anyone living more than 200 years. But for all practical purposes, science is the best we can do, and the lack of absolute confidence is what makes science so much fun.

The theory of rational self-interest escapes definitional tautology because it excludes certain interpretations and, more importantly, excludes certain observations.

Ontologically, rational self-interest says, among other things, that we do not behave like traditional computer programs or Sphex wasps, nor, more importantly, do we behave like "Kantian" moralists. Empirically, the theory of rational self-interest says that we would see different observations if people did behave like traditional computer programs or Kantian moralists.

A traditional computer program — even running on an imagined computer with as much processing power as a human brain — does not act from rational self-interest. Let's say we program a very powerful computer to actively manage some horrendously complicated non-sentient physical process, such as turbulent flow. There's no way we can independently determine a computer's subjective satisfaction: we cannot ask it if is happy or sad. We can look at all the code and find nothing that can even remotely be described as subjective satisfaction. If a program does something odd, we can find the precise cause: Some line of code does the odd thing under specific circumstances. If we change only what the code does, not what it "wants to feel", we'll change the behavior of the program.

(I'm not saying that we couldn't program a computer to act from rational self-interest; I believe that's entirely possible, hence my qualification above of traditional computer programs. But we could tell both by looking at the source code as well as empirically examining the computer's behavior that it was the sort of program that acted from rational self-interest, rather than acting "sphexishly".)

More importantly, the theory of rational self-interest says that we do not act like "Kantian" moralists*. A "Kantian" moralist is someone who acts because some act is objectively good, without regard to any notion of the actor's own subjective satisfaction. If we "feel better" for doing the objectively correct thing, that's merely a happy accident. In just the same sense, we accelerate towards the center of the Earth at 10 m/s², without regard to any notion of subjective satisfaction. If we "feel better" when we experience gravity — hardly surprising as we have evolved on a world where gravity is ubiquitous — that's merely a happy accident.

*Kant was a subtle (perhaps over-subtle) philosopher, and his moral theory is more sophisticated than the caricature that a lot of people, including many philosophers, attribute to him. But still, the trope is useful.

The difference between "Kantian" morality and gravity is of course that we accelerate towards the center of the Earth even in those circumstances where we absolutely do not want to, where there's no correlation whatsoever between what we desire and how we behave when we're falling. And we do so even when we have sufficient time to deliberate and make an informed choice about whether or not to fall.

On the other hand, people never do what's right under those circumstances when they don't have some independently identifiable subjective feeling underlying the behavior. Even when a person acts against her immediate material self-interest — when, for example, they go back into a store to pay for a missed item — you'll always hear, "I would have felt bad if I hadn't." Similarly, when someone doesn't act according to what we consider to be an important moral principle, they usually don't have an elaborate philosophical account about why that principle is incorrect, they just don't personally care, they don't feel bad about contravening the principle.

These observations are true even of people who call themselves "Kantian" moralists, of people who say they do things because it's the "right" thing to do. Even among self-described "Kantian" moralists, there's always a perfect correlation between what they consider the "right" thing do and how they subjectively feel about doing the right thing.

Indeed it is the "Kantian" moralists who are guilty of definitional tautology: if it always feels good to do the "right" thing, then in what sense is doing the right thing independent of one's subjective state? How do they account for people who not to feel good doing the "right" thing?

Of course, all of the above is not to say that our subjective feelings are not enormously complicated, and operate on many levels of abstraction. It is simply to say that we can indeed independently determine how a person feels, even in complex, abstract situations, and how they feel is the best predictor of their behavior.