Most liberals adopt some version of the Rawlsian principle that changes in the distribution of goods ought to benefit the least advantaged. This is a principle easily adopted on utilitarian grounds: the marginal utility to be derived from redistribution is quite high. For instance, the difference between the pleasure I currently experience and the pleasure I will experience with a $500 iPad is much smaller than the difference in pain alleviation that $500 can supply to a homeless person. Yet Rawls himself rejects the utilitarian justification for the maximin principle, arguing that this might sometimes justify slavery, i.e. trade offs between liberty and utility.
Like many liberals, Rawls takes the intuition about the declining marginal utility of goods back to utilitarianism itself, arguing that the same maximin principle can be adopted in the distribution of esteem and even utility. Intuitively, it seems that if we had the capacity to distribute raw utility, we ought to first distribute it to those who are, say, exceedingly depressed before distributing it to an unusually happy person. This used to be my own position.
A colleague disagreed, and pointed me to this paper by Joshua Greene and Jonathan Baron. Here’s the abstract:
In two studies, subjects judged the desirability of distributions of life expectancy or money. Their judgments showed declining marginal utility. That is, they were less sensitive to changes at the high end of each scale. Subjects also made utility ratings of the outcomes of individuals. And they made ratings of the distributions when these were described in terms of utility ratings rather than goods (years or dollars). The judgments of utility ratings showed equivalent declining marginal utility, even though they were based on utilities that themselves declined marginally. People extend their intuition about declining marginal utility to utility itself, as if utility had utility that declined marginally. In one experiment, a similar result was found with gambles: people are risk averse for utility as well as for money. We argue that this is an overextension of a reasonable heuristic and that this heuristic may account for one classic objection to utilitarian distributions.
In short, treating utility distributions the same way that we treat goods distributions isn’t rational: it’s a bias that comes from “objectifying” utility as units of “util,” and then imagining how much a person might enjoy his utils rather than seeing the utils as a the measure of that enjoyment. One interpretation of this claim, especially the analysis which calls this an “overextension,” is the following claim:
- We ought to seek Kaldor-Hicks efficiency rather than Pareto optimality in utility calculations. Unlike economic redistribution, there is no reason to prefer an equal utility gain in a low-utility individual to the same utility gain in a high-utility individual. The goal really is to maximise utility, even if that utility will not be distributed in an egalitarian fashion. (Of course there may be a fairly low limit on positive utility, which is why he’s an egalitarian.)
I tend to think this can’t be right, but consider this: it’s precisely the difference in marginal utility gains that cause us to prefer the (economically) least advantaged, so we can’t adopt a meta-maximin principle without queering the results. The problem here is that we’d otherwise end up devoting all utility gains to those who are very depressed.
In a sense, they’d become the negative version of what Nozick described as the “utility monster.” We can’t really imagine the true utility monster, someone who just reaps massive utility benefits from things ordinary people only enjoy slightly, and so ought to hoard physical goods or experiences because they enjoy them optimally. One way to get around the utility monster problem is to hypothesize a fairly low “upper-bound” on utility: there’s only so much happy a person can sustain. The declining marginal utility of more wealth and good suggests that somewhere there is an asymptotic limit to our utility. This is born out by research that demonstrates that income increases above $75,000/year don’t really increase happiness very much. Yet we can imagine the reverse fairly well: someone who is in intense pain and needs constant time and attention to feel a little bit better.
Consider borderline personality disorder. From my understanding of the underlying mechanism*, these are people who are basically really, really, sensitive. They’re incapable of separating the current sentence coming out of your mouth from the one before or after, so if that one sentence, hell, if that one word is critical, they’re feeling criticized and abused. It doesn’t matter if you’ve prefaced the criticism with a ton of praise, the moment is the only thing they’re capable of. People with BPD are like perfect Buddhists: they let go of the past and focus on the present with such desperate attention that it’s the only thing they can feel. This isn’t all bad: if it’s a nice moment, they feel like they’re on top of the world. They become addicted to that, probably in something like the technical sense of addiction. They become increasingly manipulative, constantly trying to control what’s happening right now, and to keep the stream of comfort and praise and love and attention coming.
Everyone’s a little bit like this: we all would prefer people be nice rather than mean. But most people can handle constructive criticism, can separate the particular moment from the general relationship, and feel basically whole and accepted even in the bad moments. People with BPD can’t. In that sense, their fears and vulnerabilities are completely realized in the reactions of the people around them. They’re really not likeable, insofar as anyone who spends enough time around them will find them a roller-coaster of emotions, most of which are unhealthy. Others feel like extras in the extremely selfish personal drama narrated by the disorder, and if we honestly adopted maximin principles when it came to utility, we’d be right to take on that supporting role.
Until additional gains are wasted, there’s no reason to prefer the cessation of pain in our negative utility monster to the cessation of pain in those around him. Anyone who recognizes these symptoms in their family members could probably do themselves tons of good by simply cutting their manipulative relative out of their lives, while having only a marginal impact on the unhappy state of the person with borderline personality disorder.
Yet I tend to think we do have a responsibility to devote many (economic) resources to relieving psychic pain, especially if it can be done through innovations in pharmaceutical research. If a large initial outlay allows us to relieve psychic pain using a low marginal cost pill or injection, we ought to do so. The problem with BPD is that it has thus-far shown itself to be intransigent against both drug and behavioral treatments. So when I contrast the 25,000 children who die each day from easily-treated poverty-related diseases with the massive expense and uncertain results of psychiatric drug trials, I suspect we’re still misallocating. In both cases, we’re not concerned the hypothesized upper-limit on utility, but rather with the seemingly unbounded lower reaches of pain and suffering.