Preference and utility

Maxine Udall

attempt[s] to draw a distinction between utility maximization, which is what economists have tended to assume best describes consumer behavior, and preference maximization, which is more likely to be what consumers are actually doing when markets fail, information is faulty, prices are distorted away from marginal social cost, or uncertainty (i.e., risk) cannot be or is not managed efficiently. [Economists] have tended to use the two terms as synonymous. I think the distinction is important. If preferences do not map accurately to something that can objectively be called utility, there is no guarantee that market allocations are efficient in any meaningful sense of that term.

She means this question not in its indirect, instrumental sense, but in its direct, primary sense: are there preferences we in some sense ought to have:

[John Stuart] Mill implies that there is some objective state of the world that corresponds or “should” correspond to human happiness. Were we rational, it is the state of the world we would prefer to that characterized by satisfaction of irrational, short-term pleasures.

Udall refers to Mill's famous quotation

It is better to be a human being dissatisfied than a pig satisfied; better to be Socrates dissatisfied than a fool satisfied. And if the fool, or the pig, are a different opinion, it is because they only know their own side of the question. The other party to the comparison knows both sides.

I think Udall is not correct. Trying to figure out what any philosopher "really" meant is at best an exercise for scholars (and at worst an exercise in futility); I'm neither a scholar nor a mind-reader, so I won't speculate on what Mill was thinking. It's useful, though, to apply an "epistemic test" to assertions: if you are going to assert the truth or falsity of a proposition, we must ask: how do you know — or how in principle would you know [see comments] — that it's true or false?

Mill's assertion might or might not be ontologically true, but is it epistemically determinable? Can we — even in principle — know it's true? If we cannot — even in principle — know it to be true or false, affirmation or denial of Mill's assertion becomes (at best) a religious statement. But there's no substantive ambiguity, vagueness or epistemic uncertainty about what Mill is comparing*. Mill is not saying, for example, that it's better to be an unhappy diamond than a happy emerald: even if it were the case that diamonds and emeralds secretly could be happy or unhappy, we have no way of telling the difference.

*Mill does not go to great lengths precisely defining his terms. We could probably write several graduate theses in philosophy investigating all the plausible meanings of his terms, and find that some combinations make no sense and some sensible combinations are clearly false or implausible. (And that's if we're honest philosophers; the dishonest and perverse philosophers will assume Mill is authoritative and draw false or improbable conclusions.) The point, however, is that an ordinarily imaginative person can, I think, come up with an interpretation that does make sense, and plausibly at least appears to be true.

We can unobjectionably know that Socrates is in some sense "unsatisfied" and the fool "satisfied" by ordinary observation and scientific inference. Because our inferences specifically refer to Socrates' and the fool's mental states, "satisfied" and "unsatisfied" are therefore subjective. Furthermore, everything else being equal, being "satisfied" is better than being unsatisfied. Mill, of course, is explicitly not holding everything else equal: he's drawing a distinction by comparing Socrates to the fool — another distinction we can unobjectionably know — and his "better" clearly applies to this distinction.

Granting that Mill is not making a vacuous comparison, Mill's comparison apparently cannot be a subjective comparison. It's obvious (or at least we're we're uncontroversially granting the premise) that satisfied is subjectively better than unsatisfied. Since we're making a comparison contrary to the obvious subjective comparison, to avoid contradiction it must be objective, n'est-ce pas?

The epistemic question, though, is how do or would we know that it's better to be Socrates unsatisfied than a fool satisfied? Are we using a epistemic methodology that pertains to subjective truths (truths about people's mental states) or objective truths (truths about the physical world other than of people's mental states)?

We could, for example, ask Socrates.
"We know, Socrates," we might say, "that you're dissatisfied."

"That is most assuredly true," he might reply.

"And that man over there, he is a fool and moreover completely satisfied."

"There cannot be any doubt you are again correct."

"And yet, if you could lose all your knowledge and wisdom, and become the fool, you would not want to do so, even if thereby you would become dissatisfied."

"I must agree without reservation!"

Likewise, if we asked the fool, Mill assumes he would not prefer to become wise at the cost of losing his satisfaction. Both answers seem unobjectionably plausible. But if we try to know whether it's better to be Socrates unsatisfied by just asking and accepting the answers as authoritative, we are inquiring into their mental states; we are determining their subjective preferences. Socrates prefers to be wise; the fool prefers to be ignorant. To simply say that Socrates' answer is "objectively" authoritative because he is wise would just beg the question*. (Mill does not, of course, fall into this obvious fallacy: he draws an objective distinction between Socrates and the fool: Socrates in some unobjectionable sense knows "what it's like" to be both foolish and wise, and is in a position to make the comparison; the fool, however, does not know "what it's like" to be wise.) The key here is accepting even Socrates' answer as authoritative: the mere utterance of the statement is in itself evidence of some truth. (Some truth: a statement is not authoritative evidence that the specific content of the statement is true).

*In much the same sense, Christian apologists assert — correctly, I suppose — that if you actually believe in God, then you will prefer to believe in God. The question-begging is a little more obvious in this case.

We cannot directly observe mental states. When we make assertions about a person's mental states, we are engaging in a scientific process [see comments]: First, we have a general model: people have mental states of various content, and those mental states cause them to act (considering speaking as a speech act). Their actions (including utterances) are evidentially authoritative: any theory about their specific mental states must necessarily account for all their actions. If our theory fails to explain some action, we cannot dismiss the action as irrelevant; we must change the theory. (Note that authoritative (i.e. required to be admitted to evidence) does not mean veridical (the content accurately describes reality). If a person says, "I don't like ice cream," and we later observe him eating ice cream, we might infer his statement was not veridical, but we cannot infer that his statement is not authoritative and can be ignored. We have to update our theory of his mental states to account for why he would lie. But he still in fact did lie, we need to explain the lie, and the statement is still authoritative.

We ask Socrates if he is dissatisfied, and he agrees. We ask him if the fool is satisfied; again he agrees. We ask him, "Will you take the blue pill and become a satisfied fool?" He refuses! "It's better to be wise and dissatisfied."

We must infer a set of mental states from these actions. He might, of course, simply be lying: He might not be dissatisfied in any sense; he might not actually think the fool is satisfied, or he might be lying about not wanting to take the blue pill. But this seems like an over-elaborate theory: We have to theorize not only that he's lying, but that he would want to lie. If, for example, he wanted to take the blue pill in secret, he knows (and we know, and he knows we know, etc.) that we would quickly discover he had taken the blue pill and his secret would be exposed.

Is there an alternative interpretation that's simpler? Generally, if a person acts, we infer that the action satisfies some preference; if a person refuses to act, we infer they prefer the status quo to the outcome of the action. We thus infer that Socrates prefers to be "dissatisfied", i.e. he is in one sense satisfied that he is in another sense dissatisfied.

It's plausible that "satisfied" can have a range of meanings, especially when translated to logically rigorous terminology: many words and phrases have this property. Furthermore, we also know in general that the human emotional apparatus is very complicated and multi-layered: People can have preferences about the real world (including their own bodies; they can also have preferences about preferences ("meta-preferences"), preferences about other people's preferences and their satisfaction, preferences conditioned on other peoples preferences ("If you want to see Die Hard I'd be happy to join you, but if you don't want to see it, I wouldn't go on my own account") and so forth.

So Mill's statement seems to draw a comparison contrary to only a superficial or first-level mental state; there might be meta-states — just as subjective as first-level states — that explain the discrepancy.

If we want, on the other hand, to establish an objective truth, we cannot take the mere utterance of a statement to itself be authoritative. Even the wisest person can be mistaken on a matter of objective truth: one sense of "objective truth" is a truth a wise person can be mistaken about. Whether or not Isaac Newton says that light is a particle, for example, has nothing at all to do with whether light actually is a particle. We can simply dismiss his statement as irrelevant: we don't necessarily* need our theory of light to explain why Isaac Newton said it was a particle.

*Given that Isaac Newton knew a thing or two about physics in general and optics in particular, we might want to explain his statement, but we don't have to. If, however, he argued from evidence that light was a particle, we would have to consider the evidence he used in his argument to be authoritative.

To establish an objective truth, we must establish a methodology that even a fool can follow and come to the correct answer. Indeed, we call Socrates wise in the first place precisely because he can instruct us fools into wisdom... and we want him to do so.

In a similar sense, one does not need to be a physicist to understand that objects in a gravitational field accelerate at a constant rate regardless of their mass. Rather one becomes a physicist by understanding (among other things) how objects behave in gravity. If we are going to say the fool's preference to remain foolish is mistaken, we must also say we can somehow instruct the fool to the truth. If we say a person is a fool because he prefers to remain foolish, we are again begging the question.

We can persuade the "fool" such as Newton who believes light is a particle that it is really a wave* by showing him the evidence. Is there evidence we can show the satisfied fool that he should in any sense prefer to be a dissatisfied philosopher? We could, of course, make him a philosopher and ask him again (and presumably he would answer as Socrates), but this "solution" just begs the question. We have "magically" changed his preferences and mirable dictu! we discover his preferences have changed. Not the most compelling argument, but I don't think there's a better one. Indeed, Mill's statement rings true precisely because people generally fell satisfied at an abstract meta-level to be as wise as they are, even at the cost of superficially or low-level dissatisfaction that would go away if they were less wise.

*Yeah, yeah, I know: quantum mechanics.

In short, even if we accept Mill's statement as definitely true in some sense, we cannot simply infer that there are actual objectively true statements about the preferences we should have. The explanation of meta-preferences and the general complexity and multi-level account of preferences, emotions and mental states (well-supported by independent evidence) adequately explains Mill's statement.